ifeq "$(USE_FRONTEND)" "1"
# common
-OBJS += platform/common/main.o platform/common/emu.o \
+OBJS += platform/common/main.o platform/common/emu.o platform/common/upscale.o \
platform/common/menu_pico.o platform/common/config_file.o
# libpicofe
FinalizeLine555(sh, line, est);
if ((Pico32x.vdp_regs[0] & P32XV_Mx) == 0 || // 32x blanking
- // XXX: how is 32col mode hadled by real hardware?
+ // XXX: how is 32col mode handled by real hardware?
!(Pico.video.reg[12] & 1) || // 32col mode
(Pico.video.debug_p & PVD_KILL_32X))
{
#define PICOSCAN_POST \
PicoScan32xEnd(l + (lines_sft_offs & 0xff)); \
+ Pico.est.DrawLineDest = (char *)Pico.est.DrawLineDest + DrawLineDestIncrement32x; \
#define make_do_loop(name, pre_code, post_code, md_code) \
/* Direct Color Mode */ \
int which_func;
Pico.est.DrawLineDest = (char *)DrawLineDestBase32x + offs * DrawLineDestIncrement32x;
+ Pico.est.DrawLineDestIncr = DrawLineDestIncrement32x;
dram = Pico32xMem->dram[Pico32x.vdp_regs[0x0a/2] & P32XV_FS];
if (Pico32xDrawMode == PDM32X_BOTH)
dst[p + 2] = pal[*pmd++];
dst[p + 3] = pal[*pmd++];
}
- dst = (void *)((char *)dst + DrawLineDestIncrement32x);
+ dst = Pico.est.DrawLineDest = (char *)dst + DrawLineDestIncrement32x;
pmd += 328 - plen;
if (have_scan)
PicoScan32xEnd(l + offs);
PIC_LDR_INIT()
-.macro call_scan_prep cond est @ &Pico.est
+.macro call_scan_prep cond pico @ &Pico
.if \cond
PIC_LDR(r4, r6, PicoScan32xBegin)
PIC_LDR(r5, r6, PicoScan32xEnd)
- ldr r6, [\est, #OFS_EST_DrawLineDest]
+ add r6, \pico, #OFS_Pico_est
ldr r4, [r4]
ldr r5, [r5]
stmfd sp!, {r4,r5,r6}
add r0, r0, r4
mov lr, pc
ldr pc, [sp, #(3+0)*4]
- ldr r0, [sp, #(3+2)*4] @ &DrawLineDest
+ ldr r0, [sp, #(3+2)*4] @ &Pico.est
ldmfd sp!, {r1-r3}
- ldr r0, [r0]
+ ldr r0, [r0, #OFS_EST_DrawLineDest]
.endif
.endm
add r0, r0, r4
mov lr, pc
ldr pc, [sp, #(4+1)*4]
+ ldr r1, [sp, #(3+2)*4] @ &Pico.est
+ ldr r0, [r1, #OFS_EST_DrawLineDest]
+ ldr r2, [r1, #OFS_EST_DrawLineDestIncr]
+ add r0, r0, r2
+ str r0, [r1]
ldmfd sp!, {r0-r3}
.endif
.endm
PIC_LDR(lr, r9, Pico)
PIC_LDR(r10,r9, Pico32x)
ldr r11, [lr, #OFS_Pico_est+OFS_EST_Draw2FB]
+ ldr r12, [lr, #OFS_Pico_est+OFS_EST_DrawLineDestIncr]
ldrh r10,[r10, #0x40] @ Pico32x.vdp_regs[0]
add r9, lr, #OFS_Pico_est+OFS_EST_HighPal @ palmd
0: @ loop_outer:
call_scan_end \call_scan
+ ldr r12, [r9, #OFS_EST_DrawLineDestIncr-OFS_EST_HighPal]
+ sub r0, r0, #320*2
+ add r0, r0, r12
add r4, r4, #1
cmp r4, r2, lsr #16
call_scan_fin_ge \call_scan
PIC_LDR(lr, r9, Pico)
PIC_LDR(r10,r9, Pico32xMem)
- ldr r9,=OFS_PMEM32x_pal_native
+ ldr r9, =OFS_PMEM32x_pal_native
ldr r10, [r10]
ldr r11, [lr, #OFS_Pico_est+OFS_EST_Draw2FB]
add r10,r10,r9
0: @ loop_outer:
call_scan_end \call_scan
+ ldr r12, [r9, #OFS_EST_DrawLineDestIncr-OFS_EST_HighPal]
+ sub r0, r0, #320*2
+ add r0, r0, r12
add r4, r4, #1
cmp r4, r2, lsr #16
call_scan_fin_ge \call_scan
PIC_LDR(lr, r9, Pico)
PIC_LDR(r10,r9, Pico32xMem)
- ldr r9,=OFS_PMEM32x_pal_native
+ ldr r9, =OFS_PMEM32x_pal_native
ldr r10, [r10]
ldr r11, [lr, #OFS_Pico_est+OFS_EST_Draw2FB]
add r10,r10,r9
0: @ loop_outer:
call_scan_end \call_scan
+ ldr r12, [r9, #OFS_EST_DrawLineDestIncr-OFS_EST_HighPal]
+ sub r0, r0, #320*2
+ add r0, r0, r12
add r4, r4, #1
cmp r4, r2, lsr #16
call_scan_fin_ge \call_scan
*/\r
\r
#include "pico_int.h"\r
+#include <platform/common/upscale.h>\r
+\r
#define FORCE // layer forcing via debug register?\r
\r
int (*PicoScanBegin)(unsigned int num) = NULL;\r
\r
if (!(Pico.video.reg[12]&1))\r
max_sprites = 64, max_line_sprites = 16, max_width = 264;\r
- if (PicoIn.opt & POPT_DIS_SPRITE_LIM)\r
+ if (*est->PicoOpt & POPT_DIS_SPRITE_LIM)\r
max_line_sprites = MAX_LINE_SPRITES;\r
\r
sh = Pico.video.reg[0xC]&8; // shadow/hilight?\r
if (Pico.video.reg[12]&1) {\r
len = 320;\r
} else {\r
- if (!(PicoIn.opt&POPT_DIS_32C_BORDER)) pd+=32;\r
len = 256;\r
}\r
\r
- {\r
-#if 1\r
- int i;\r
-\r
- for (i = len; i > 0; i-=4) {\r
- *pd++ = pal[*ps++];\r
- *pd++ = pal[*ps++];\r
- *pd++ = pal[*ps++];\r
- *pd++ = pal[*ps++];\r
+ if ((*est->PicoOpt & POPT_EN_SOFTSCALE) && len == 256) {\r
+ switch (PicoIn.filter) {\r
+ case 3: h_upscale_bl4_4_5(pd, 320, ps, 256, 256, f_pal); break;\r
+ case 2: h_upscale_bl2_4_5(pd, 320, ps, 256, 256, f_pal); break;\r
+ case 1: h_upscale_snn_4_5(pd, 320, ps, 256, 256, f_pal); break;\r
+ default: h_upscale_nn_4_5(pd, 320, ps, 256, 256, f_pal); break;\r
}\r
+ } else {\r
+ if (!(*est->PicoOpt & POPT_DIS_32C_BORDER) && len == 256) pd += 32;\r
+#if 1\r
+ h_copy(pd, 320, ps, 320, len, f_pal);\r
#else\r
extern void amips_clut(unsigned short *dst, unsigned char *src, unsigned short *pal, int count);\r
extern void amips_clut_6bit(unsigned short *dst, unsigned char *src, unsigned short *pal, int count);\r
if (Pico.video.reg[12]&1) {\r
len = 320;\r
} else {\r
- if (!(PicoIn.opt & POPT_DIS_32C_BORDER))\r
- pd += 32;\r
len = 256;\r
}\r
\r
- if (DrawLineDestIncrement == 0) {\r
+ if ((PicoIn.opt & POPT_EN_SOFTSCALE) && len == 256) {\r
+ unsigned char *ps = est->HighCol+8;\r
+ unsigned char pal = 0;\r
+\r
+ if (!sh && (est->rendstatus & PDRAW_SONIC_MODE))\r
+ pal = est->SonicPalCount*0x40;\r
+ if (DrawLineDestIncrement == 0)\r
+ pd = est->HighCol+8;\r
+ // Smoothing can't be used with CLUT, hence it's always Nearest Neighbour.\r
+ // use reverse version since src and dest ptr may be the same.\r
+ rh_upscale_nn_4_5(pd, 320, ps, 256, len, f_or);\r
+ } else if (DrawLineDestIncrement == 0) {\r
if (!sh && (est->rendstatus & PDRAW_SONIC_MODE))\r
- blockcpy_or(pd+8, est->HighCol+8, len, est->SonicPalCount*0x40);\r
- } else if (!sh && (est->rendstatus & PDRAW_SONIC_MODE)) {\r
- // select active backup palette\r
- blockcpy_or(pd, est->HighCol+8, len, est->SonicPalCount*0x40);\r
+ blockcpy_or(est->HighCol+8, est->HighCol+8, len, est->SonicPalCount*0x40);\r
} else {\r
- blockcpy(pd, est->HighCol+8, len);\r
+ if (!(PicoIn.opt & POPT_DIS_32C_BORDER))\r
+ pd += 32;\r
+ if (!sh && (est->rendstatus & PDRAW_SONIC_MODE)) {\r
+ // select active backup palette\r
+ blockcpy_or(pd, est->HighCol+8, len, est->SonicPalCount*0x40);\r
+ } else {\r
+ blockcpy(pd, est->HighCol+8, len);\r
+ }\r
}\r
}\r
\r
// MUST be called every frame\r
PICO_INTERNAL void PicoFrameStart(void)\r
{\r
- int offs = 8, lines = 224;\r
+ int loffs = 8, lines = 224, coffs = 0, columns = 320;\r
int dirty = ((Pico.est.rendstatus & PDRAW_SONIC_MODE) || Pico.m.dirtyPal);\r
int sprep = Pico.est.rendstatus & (PDRAW_SPRITES_MOVED|PDRAW_DIRTY_SPRITES);\r
int skipped = Pico.est.rendstatus & PDRAW_SKIP_FRAME;\r
Pico.est.rendstatus = 0;\r
if ((Pico.video.reg[12] & 6) == 6)\r
Pico.est.rendstatus |= PDRAW_INTERLACE; // interlace mode\r
- if (!(Pico.video.reg[12] & 1))\r
+ if (!(Pico.video.reg[12] & 1)) {\r
Pico.est.rendstatus |= PDRAW_32_COLS;\r
+ if (!(PicoIn.opt & POPT_EN_SOFTSCALE)) {\r
+ columns = 256;\r
+ coffs = 32;\r
+ }\r
+ }\r
if (Pico.video.reg[1] & 8) {\r
Pico.est.rendstatus |= PDRAW_30_ROWS;\r
- offs = 0;\r
lines = 240;\r
+ loffs = 0;\r
}\r
+ if (PicoIn.opt & POPT_DIS_32C_BORDER)\r
+ coffs = 0;\r
\r
if (Pico.est.rendstatus != rendstatus_old || lines != rendlines) {\r
rendlines = lines;\r
// mode_change() might reset rendstatus_old by calling SetColorFormat\r
- emu_video_mode_change((lines == 240) ? 0 : 8,\r
- lines, (Pico.video.reg[12] & 1) ? 0 : 1);\r
+ emu_video_mode_change(loffs, lines, coffs, columns);\r
rendstatus_old = Pico.est.rendstatus;\r
}\r
if (PicoIn.skipFrame) // preserve this until something is rendered at last\r
if (sprep | skipped)\r
Pico.est.rendstatus |= PDRAW_PARSE_SPRITES;\r
\r
- Pico.est.HighCol = HighColBase + offs * HighColIncrement;\r
- Pico.est.DrawLineDest = (char *)DrawLineDestBase + offs * DrawLineDestIncrement;\r
+ Pico.est.HighCol = HighColBase + loffs * HighColIncrement;\r
+ Pico.est.DrawLineDest = (char *)DrawLineDestBase + loffs * DrawLineDestIncrement;\r
Pico.est.DrawScanline = 0;\r
skip_next_line = 0;\r
\r
tst r12, #1\r
movne r2, #320/8 @ len\r
bne .fl_no32colRGB555\r
- ldr r4, [r10, #OFS_EST_PicoOpt]\r
+ ldr r5, [r10, #OFS_EST_PicoOpt]\r
mov r2, #256/8\r
- ldr r4, [r4]\r
+ ldr r4, [r5]\r
tst r4, #0x4000\r
bne .fl_32scale_RGB555\r
tst r4, #0x0100\r
\r
\r
.fl_32scale_RGB555:\r
- mov r9, #0x3900 @ f800 07e0 001f | e000 0780 001c | 3800 01e0 0007\r
- orr r9, r9, #0x00e7\r
+ ldr r5, [r5, #OFS_PicoIn_filter-OFS_PicoIn_opt]\r
+\r
+ mov r9, #0xf700 @ f800 07e0 001f | e000 0780 001c | 3800 01e0 0007\r
+ orr r9, r9, #0x00de\r
\r
#ifdef UNALIGNED_DRAWLINEDEST\r
tst r0, #2\r
bne .fl_32scale_RGB555u\r
#endif\r
\r
-.fl_loop32scale_RGB555:\r
+ ands r5, r5, #0x3\r
+ addne pc, pc, r5, lsl #2\r
+ b .fl_32scale_nn\r
+ b .fl_32scale_nn\r
+ b .fl_32scale_snn\r
+ b .fl_32scale_bl2\r
+ b .fl_32scale_bl4\r
+\r
+.fl_32scale_nn:\r
+ ldr r12, [r1], #4\r
+ ldr r7, [r1], #4\r
+\r
+ and r4, lr, r12, lsl #1\r
+ ldrh r4, [r3, r4]\r
+ and r5, lr, r12, lsr #7\r
+ ldrh r5, [r3, r5]\r
+ and r6, lr, r12, lsr #15\r
+ ldrh r6, [r3, r6]\r
+ and r10,lr, r12, lsr #23\r
+ ldrh r10,[r3, r10]\r
+\r
+ orr r4, r4, r5, lsl #16\r
+ orr r5, r6, r6, lsl #16\r
+\r
+ and r6, lr, r7, lsl #1\r
+ ldrh r6, [r3, r6]\r
+ and r8, lr, r7, lsr #7\r
+ ldrh r8, [r3, r8]\r
+ and r12,lr, r7, lsr #15\r
+ ldrh r12,[r3, r12]\r
+ and r7, lr, r7, lsr #23\r
+ ldrh r7, [r3, r7]\r
+\r
+ orr r6, r10,r6, lsl #16\r
+ orr r8, r8,r12, lsl #16\r
+\r
+ subs r2, r2, #1\r
+\r
+ orr r10,r12,r7, lsl #16\r
+\r
+ stmia r0!, {r4,r5,r6,r8,r10}\r
+ bne .fl_32scale_nn\r
+\r
+ ldmfd sp!, {r4-r10,pc}\r
+\r
+.fl_32scale_snn:\r
+ ldr r12, [r1], #4\r
+ ldr r7, [r1], #4\r
+\r
+ and r4, lr, r12, lsl #1\r
+ ldrh r4, [r3, r4]\r
+ and r5, lr, r12, lsr #7\r
+ ldrh r5, [r3, r5]\r
+ and r6, lr, r12, lsr #15\r
+ ldrh r6, [r3, r6]\r
+ and r10,lr, r12, lsr #23\r
+ ldrh r10,[r3, r10]\r
+\r
+ and r4, r4, r9\r
+ and r5, r5, r9\r
+ orr r4, r4, r5, lsl #16\r
+ and r6, r6, r9\r
+ add r5, r5, r6\r
+ mov r5, r5, lsr #1\r
+ orr r5, r5, r6, lsl #16\r
+\r
+ and r6, lr, r7, lsl #1\r
+ ldrh r6, [r3, r6]\r
+ and r8, lr, r7, lsr #7\r
+ ldrh r8, [r3, r8]\r
+ and r12,lr, r7, lsr #15\r
+ ldrh r12,[r3, r12]\r
+ and r7, lr, r7, lsr #23\r
+ ldrh r7, [r3, r7]\r
+\r
+ and r6, r6, r9\r
+ and r10,r10,r9\r
+ orr r6, r10,r6, lsl #16\r
+ and r12,r12,r9\r
+ and r7, r7, r9\r
+ orr r10,r12,r7, lsl #16\r
+\r
+ and r8, r8, r9\r
+ add r12,r12,r8\r
+ mov r12,r12,lsr #1\r
+ orr r8, r8,r12, lsl #16\r
+\r
+ subs r2, r2, #1\r
+\r
+ stmia r0!, {r4,r5,r6,r8,r10}\r
+ bne .fl_32scale_snn\r
+\r
+ ldmfd sp!, {r4-r10,pc}\r
+\r
+.fl_32scale_bl2:\r
+ ldr r12, [r1], #4\r
+ ldr r7, [r1], #4\r
+\r
+ and r4, lr, r12, lsl #1\r
+ ldrh r4, [r3, r4]\r
+ and r5, lr, r12, lsr #7\r
+ ldrh r5, [r3, r5]\r
+ and r6, lr, r12, lsr #15\r
+ ldrh r6, [r3, r6]\r
+\r
+ and r4, r4, r9\r
+ and r5, r5, r9\r
+ add r10,r4, r5\r
+ mov r10,r10,lsr #1\r
+ orr r4, r4, r10,lsl #16 @ px0 | (px0+px1)/2\r
+\r
+ and r6, r6, r9\r
+ add r5, r5, r6\r
+ mov r5, r5, lsr #1\r
+ orr r5, r5, r6, lsl #16 @ (px1+px2)/2 | px2\r
+\r
+ and r10,lr, r12, lsr #23\r
+ ldrh r10,[r3, r10]\r
+ and r8, lr, r7, lsl #1\r
+ ldrh r8, [r3, r8]\r
+\r
+ and r10,r10,r9\r
+ and r8, r8, r9\r
+ orr r6, r10,r8, lsl #16 @ px3 | px4\r
+\r
+ and r12,lr, r7, lsr #15\r
+ ldrh r12,[r3, r12]\r
+ and r10, lr, r7, lsr #23\r
+ ldrh r10, [r3, r10]\r
+ and r7, lr, r7, lsr #7\r
+ ldrh r7, [r3, r7]\r
+\r
+ and r12,r12,r9\r
+ and r10,r10,r9\r
+ orr r10,r12,r10, lsl #16 @ px6 | px7\r
+\r
+ and r7, r7, r9\r
+ add r12,r12,r7\r
+ add r8, r8, r7\r
+ mov r8, r8, lsr #1\r
+ mov r12,r12,lsr #1\r
+ orr r8, r8,r12, lsl #16 @ (px4+px5)/2 | (px5+px6)/2\r
+\r
+ subs r2, r2, #1\r
+\r
+ stmia r0!, {r4,r5,r6,r8,r10}\r
+ bne .fl_32scale_bl2\r
+\r
+ ldmfd sp!, {r4-r10,pc}\r
+\r
+.fl_32scale_bl4:\r
ldr r12, [r1], #4\r
ldr r7, [r1], #4\r
\r
ldrh r4, [r3, r4]\r
and r5, lr, r12,lsr #7\r
ldrh r5, [r3, r5]\r
- and r4, r4, r9, lsl #2\r
+\r
+ @ r4 = 1/4px0+3/4px1 : px0\r
+ and r4, r4, r9\r
orr r4, r4, r4, lsl #14 @ r4[31:16] = 1/4 pix_s 0\r
- and r5, r5, r9, lsl #2\r
+ and r5, r5, r9\r
sub r6, r5, r5, lsr #2 @ r6 = 3/4 pix_s 1\r
add r4, r4, r6, lsl #16 @ pix_d 0, 1\r
+\r
and r6, lr, r12,lsr #15\r
ldrh r6, [r3, r6]\r
and r12,lr, r12,lsr #23\r
ldrh r12,[r3, r12]\r
- and r6, r6, r9, lsl #2\r
+\r
+ @ r5 = 3/4px2+1/4px3 : (px1+px2)/2\r
+ and r6, r6, r9\r
add r5, r5, r6\r
mov r5, r5, lsr #1\r
sub r6, r6, r6, lsr #2 @ r6 = 3/4 pix_s 2\r
\r
and r6, lr, r7, lsl #1\r
ldrh r6, [r3, r6]\r
- and r12,r12,r9, lsl #2\r
+ and r12,r12,r9\r
add r5, r5, r12,lsl #14 @ pix_d 2, 3\r
- and r6, r6, r9, lsl #2\r
+\r
+ @ r6 = px4 : px3\r
+ and r6, r6, r9\r
orr r6, r12,r6, lsl #16 @ pix_d 4, 5\r
\r
+ @ r8 = (px5+px6)/2 : 1/4px4+3/4px5\r
and r12,lr, r7, lsr #7\r
ldrh r12,[r3, r12]\r
and r10,lr, r7, lsr #15\r
ldrh r10,[r3, r10]\r
- and r12,r12,r9, lsl #2\r
+ and r12,r12,r9\r
sub r8, r12,r12,lsr #2 @ r8 = 3/4 pix_s 1\r
add r8, r8, r6, lsr #18\r
+\r
and r7, lr, r7, lsr #23\r
ldrh r7, [r3, r7]\r
- and r10,r10,r9, lsl #2\r
+ and r10,r10,r9\r
orr r8, r8, r10,lsl #15\r
add r8, r8, r12,lsl #15 @ pix_d 6, 7\r
+\r
+ @ r10 = px7 : 3/4px6+1/4px7\r
sub r10,r10,r10,lsr #2 @ r10= 3/4 pix_s 2\r
- and r7, r7, r9, lsl #2\r
+ and r7, r7, r9\r
add r10,r10,r7, lsr #2 @ += 1/4 pix_s 3\r
orr r10,r10,r7, lsl #16 @ pix_d 8, 9\r
\r
subs r2, r2, #1\r
\r
stmia r0!, {r4,r5,r6,r8,r10}\r
- bne .fl_loop32scale_RGB555\r
+ bne .fl_32scale_bl4\r
\r
ldmfd sp!, {r4-r10,pc}\r
\r
ldrh r6, [r3, r6]\r
and r5, lr, r12,lsr #7\r
ldrh r5, [r3, r5]\r
- and r6, r6, r9, lsl #2\r
+ and r6, r6, r9\r
orr r4, r4, r6, lsl #16 @ r4 = pix_d -1, 0\r
\r
- and r5, r5, r9, lsl #2\r
+ and r5, r5, r9\r
sub r8, r5, r5, lsr #2 @ r8 = 3/4 pix_s 1\r
add r6, r8, r6, lsr #2 @ r6 = (1/4 pix_s 0) + (3/4 pix_s 1)\r
orr r5, r6, r5, lsl #15\r
ldrh r6, [r3, r6]\r
and r12,lr, r12,lsr #23\r
ldrh r12,[r3, r12]\r
- and r6, r6, r9, lsl #2\r
+ and r6, r6, r9\r
add r5, r5, r6, lsl #15 @ r5 = pix_d 1, 2\r
\r
and r8, lr, r7, lsl #1\r
ldrh r8, [r3, r8]\r
and r10,lr, r7, lsr #7\r
ldrh r10,[r3, r10]\r
- and r12,r12,r9, lsl #2\r
+ and r12,r12,r9\r
sub r6, r6, r6, lsr #2 @ r6 = 3/4 pix_s 2\r
add r6, r6, r12,lsr #2\r
orr r6, r6, r12,lsl #16 @ r6 = pix_d 3, 4\r
\r
- and r8, r8, r9, lsl #2\r
- and r10,r10,r9, lsl #2\r
+ and r8, r8, r9\r
+ and r10,r10,r9\r
sub r12,r10,r10,lsr #2 @ r12 = 3/4 pix_s 5\r
orr r8, r8, r8, lsl #14\r
add r8, r8, r12,lsl #16 @ r8 = pix_d 5, 6\r
ldrh r12,[r3, r12]\r
and r7, lr, r7, lsr #23\r
ldrh r7, [r3, r7]\r
- and r12,r12,r9, lsl #2\r
+ and r12,r12,r9\r
add r10,r10,r12\r
mov r10,r10, lsr #1\r
sub r12,r12,r12,lsr #2 @ r12 = 3/4 pix_s 6\r
orr r10,r10,r12,lsl #16\r
- and r7, r7, r9, lsl #2\r
+ and r7, r7, r9\r
add r10,r10,r7, lsl #14 @ r10 = pix_d 7, 8\r
\r
subs r2, r2, #1\r
}
if (Pico.est.rendstatus != rendstatus_old || lines != rendlines) {
- emu_video_mode_change(screen_offset, lines, 1);
+ emu_video_mode_change(screen_offset, lines, line_offset, 256);
rendstatus_old = Pico.est.rendstatus;
rendlines = lines;
}
Pico.est.HighPal[0xe0] = 0;
}
+#include <platform/common/upscale.h>
+
static void FinalizeLineRGB555M4(int line)
{
if (Pico.m.dirtyPal)
static void FinalizeLine8bitM4(int line)
{
- unsigned char *pd = Pico.est.DrawLineDest;
-
- if (DrawLineDestIncrement)
- memcpy(pd + line_offset, Pico.est.HighCol + line_offset + 8, 256);
+ unsigned char *pd = Pico.est.DrawLineDest + line_offset;
+ unsigned char *ps = Pico.est.HighCol + line_offset + 8;
+
+ if (DrawLineDestIncrement) {
+ if (PicoIn.opt & POPT_EN_SOFTSCALE)
+ rh_upscale_nn_4_5(pd, 320, ps, 256, 256, f_nop);
+ else
+ memcpy(pd, ps, 256);
+ }
}
void PicoDrawSetOutputMode4(pdso_t which)
{
- line_offset = PicoIn.opt & POPT_DIS_32C_BORDER ? 0 : 32;
+ line_offset = PicoIn.opt & (POPT_DIS_32C_BORDER|POPT_EN_SOFTSCALE) ? 0 : 32;
switch (which)
{
case PDF_8BIT: FinalizeLineM4 = FinalizeLine8bitM4; break;
extern int plat_mem_set_exec(void *ptr, size_t size);\r
\r
// this one should handle display mode changes\r
-extern void emu_video_mode_change(int start_line, int line_count, int is_32cols);\r
+extern void emu_video_mode_change(int start_line, int line_count, int start_col, int col_count);\r
\r
// this must switch to 16bpp mode\r
extern void emu_32x_startup(void);\r
\r
// the emulator is configured and some status is reported\r
// through this global state (not saved in savestates)\r
-typedef struct\r
+typedef struct PicoInterface\r
{\r
unsigned int opt; // POPT_* bitfield\r
\r
unsigned short quirks; // game-specific quirks: PQUIRK_*\r
unsigned short overclockM68k; // overclock the emulated 68k, in %\r
\r
+ unsigned short filter; // softscale filter type\r
+\r
int sndRate; // rate in Hz\r
int sndFilterAlpha; // Low pass sound filter alpha (Q16)\r
short *sndOut; // PCM output buffer\r
int DrawScanline;\r
int rendstatus;\r
void *DrawLineDest; // draw destination\r
+ int DrawLineDestIncr;\r
unsigned char *HighCol;\r
s32 *HighPreSpr;\r
struct Pico *Pico;\r
memset32((short *)g_screen_ptr + g_screen_ppitch * y, 0,\r
g_screen_width * 2 / 4);\r
\r
- PicoIn.opt &= ~POPT_ALT_RENDERER;\r
+ PicoIn.opt &= ~(POPT_ALT_RENDERER|POPT_EN_SOFTSCALE);\r
PicoIn.opt |= POPT_ACC_SPRITES;\r
if (!no_scale && currentConfig.scaling)\r
PicoIn.opt |= POPT_EN_SOFTSCALE;\r
enum {
EOPT_SCALE_NONE = 0,
- EOPT_SCALE_SW,
+ // linux, GP2X:
+ EOPT_SCALE_SW = 1,
EOPT_SCALE_HW,
};
+enum {
+ EOPT_FILTER_NONE = 0,
+ // software scalers
+ EOPT_FILTER_SMOOTHER = 1,
+ EOPT_FILTER_BILINEAR1,
+ EOPT_FILTER_BILINEAR2,
+};
+
enum {
EOPT_CONFIRM_NONE = 0,
EOPT_CONFIRM_SAVE = 1,
int CPUclock;
int volume;
int gamma;
- int scaling; // gp2x: EOPT_SCALE_*; psp: bilinear filtering
+ int scaling; // EOPT_SCALE_*
int vscaling;
int rotation; // for UIQ
float scale; // psp: screen scale
int turbo_rate;
int renderer;
int renderer32x;
- int filter; // pandora
+ int filter; // EOPT_FILTER_* video filter
int analog_deadzone;
int msh2_khz;
int ssh2_khz;
/* should be in libpicofe/plat.h */
void plat_video_clear_status(void);
void plat_video_clear_buffers(void);
+void plat_video_set_size(int w, int h);
#ifdef __cplusplus
} // extern "C"
#include <pico/pico_int.h>
static void *shadow_fb;
+static struct area { int w, h; } area;
static struct in_pdata in_sdl_platform_data = {
.defbinds = in_sdl_defbinds,
}
}
-void rgb565_to_uyvy(void *d, const void *s, int pixels, int x2)
+void rgb565_to_uyvy(void *d, const void *s, int w, int h, int pitch, int x2)
{
uint32_t *dst = d;
const uint16_t *src = s;
+ int i;
- if (x2)
- for (; pixels > 0; src += 4, dst += 4, pixels -= 4)
- {
- struct uyvy *uyvy0 = yuv_uyvy + src[0], *uyvy1 = yuv_uyvy + src[1];
- struct uyvy *uyvy2 = yuv_uyvy + src[2], *uyvy3 = yuv_uyvy + src[3];
+ if (x2) while (h--) {
+ for (i = w; i > 0; src += 4, dst += 4, i -= 4)
+ {
+ struct uyvy *uyvy0 = yuv_uyvy + src[0], *uyvy1 = yuv_uyvy + src[1];
+ struct uyvy *uyvy2 = yuv_uyvy + src[2], *uyvy3 = yuv_uyvy + src[3];
#if CPU_IS_LE
- dst[0] = (uyvy0->y << 24) | uyvy0->vyu;
- dst[1] = (uyvy1->y << 24) | uyvy1->vyu;
- dst[2] = (uyvy2->y << 24) | uyvy2->vyu;
- dst[3] = (uyvy3->y << 24) | uyvy3->vyu;
+ dst[0] = (uyvy0->y << 24) | uyvy0->vyu;
+ dst[1] = (uyvy1->y << 24) | uyvy1->vyu;
+ dst[2] = (uyvy2->y << 24) | uyvy2->vyu;
+ dst[3] = (uyvy3->y << 24) | uyvy3->vyu;
#else
- dst[0] = uyvy0->y | (uyvy0->vyu << 8);
- dst[1] = uyvy1->y | (uyvy1->vyu << 8);
- dst[2] = uyvy2->y | (uyvy2->vyu << 8);
- dst[3] = uyvy3->y | (uyvy3->vyu << 8);
+ dst[0] = uyvy0->y | (uyvy0->vyu << 8);
+ dst[1] = uyvy1->y | (uyvy1->vyu << 8);
+ dst[2] = uyvy2->y | (uyvy2->vyu << 8);
+ dst[3] = uyvy3->y | (uyvy3->vyu << 8);
#endif
- } else
- for (; pixels > 0; src += 4, dst += 2, pixels -= 4)
- {
- struct uyvy *uyvy0 = yuv_uyvy + src[0], *uyvy1 = yuv_uyvy + src[1];
- struct uyvy *uyvy2 = yuv_uyvy + src[2], *uyvy3 = yuv_uyvy + src[3];
+ }
+ src += pitch - w;
+ } else while (h--) {
+ for (i = w; i > 0; src += 4, dst += 2, i -= 4)
+ {
+ struct uyvy *uyvy0 = yuv_uyvy + src[0], *uyvy1 = yuv_uyvy + src[1];
+ struct uyvy *uyvy2 = yuv_uyvy + src[2], *uyvy3 = yuv_uyvy + src[3];
#if CPU_IS_LE
- dst[0] = (uyvy1->y << 24) | uyvy0->vyu;
- dst[1] = (uyvy3->y << 24) | uyvy2->vyu;
+ dst[0] = (uyvy1->y << 24) | uyvy0->vyu;
+ dst[1] = (uyvy3->y << 24) | uyvy2->vyu;
#else
- dst[0] = uyvy1->y | (uyvy0->vyu << 8);
- dst[1] = uyvy3->y | (uyvy2->vyu << 8);
+ dst[0] = uyvy1->y | (uyvy0->vyu << 8);
+ dst[1] = uyvy3->y | (uyvy2->vyu << 8);
#endif
+ }
+ src += pitch - w;
}
}
static int clear_buf_cnt, clear_stat_cnt;
+void plat_video_set_size(int w, int h)
+{
+ if (area.w != w || area.h != h) {
+ area = (struct area) { w, h };
+
+ if (plat_sdl_change_video_mode(w, h, 0) < 0) {
+ // failed, revert to original resolution
+ plat_sdl_change_video_mode(g_screen_width, g_screen_height, 0);
+ w = g_screen_width, h = g_screen_height;
+ }
+ if (!plat_sdl_overlay && !plat_sdl_gl_active) {
+ g_screen_width = w;
+ g_screen_height = h;
+ g_screen_ppitch = w;
+ g_screen_ptr = plat_sdl_screen->pixels;
+ }
+ }
+}
+
void plat_video_flip(void)
{
if (plat_sdl_overlay != NULL) {
SDL_Rect dstrect =
{ 0, 0, plat_sdl_screen->w, plat_sdl_screen->h };
-
SDL_LockYUVOverlay(plat_sdl_overlay);
rgb565_to_uyvy(plat_sdl_overlay->pixels[0], shadow_fb,
- g_screen_ppitch * g_screen_height,
- plat_sdl_overlay->w > 2*plat_sdl_overlay->h);
+ area.w, area.h, g_screen_ppitch,
+ plat_sdl_overlay->w >= 2*area.w);
SDL_UnlockYUVOverlay(plat_sdl_overlay);
SDL_DisplayYUVOverlay(plat_sdl_overlay, &dstrect);
}
SDL_LockYUVOverlay(plat_sdl_overlay);
rgb565_to_uyvy(plat_sdl_overlay->pixels[0], shadow_fb,
- g_menuscreen_pp * g_menuscreen_h, 0);
+ g_menuscreen_w, g_menuscreen_h, g_menuscreen_pp, 0);
SDL_UnlockYUVOverlay(plat_sdl_overlay);
SDL_DisplayYUVOverlay(plat_sdl_overlay, &dstrect);
void plat_video_loop_prepare(void)
{
- // take over any new vout settings XXX ask plat_sdl for scaling instead!
+ // take over any new vout settings
plat_sdl_change_video_mode(g_menuscreen_w, g_menuscreen_h, 0);
// switch over to scaled output if available
- if (plat_sdl_overlay != NULL || plat_sdl_gl_active || currentConfig.scaling != EOPT_SCALE_NONE) {
+ if (plat_sdl_overlay != NULL || plat_sdl_gl_active) {
g_screen_width = 320;
g_screen_height = 240;
g_screen_ppitch = g_screen_width;
g_screen_ptr = plat_sdl_screen->pixels;
}
plat_video_set_buffer(g_screen_ptr);
+ plat_video_set_size(g_screen_width, g_screen_height);
}
void plat_early_init(void)
* nn: nearest neighbour
* snn: "smoothed" nearest neighbour (see below)
* bln: n-level-bilinear with n quantized weights
- * quantization: 0: a<1/2*n, 1/n: 1/2*n<=a<3/2*n, etc
+ * quantization: 0: a<1/(2*n), 1/n: 1/(2*n)<=a<3/(2*n), etc
* currently n=2, n=4 are implemented (there's n=8 mixing, but no filters)
* [NB this has been brought to my attn, which is probably the same as bl2:
* https://www.drdobbs.com/image-scaling-with-bresenham/184405045?pgno=1]
* a sharper look than a bilinear filter, at the price of some visible jags
* on diagonal edges.
*
- * scaling modes:
- * 256x___ -> 320x___ only horizontal scaling. Produces an aspect error of
- * ~7% for NTSC 224 line modes, but is correct for PAL
- * 256/320x224/240
- * -> 320x240 always produces 320x240 at DAR 4:3
- * 160x144 -> 320x240 game gear (currently unused)
+ * example scaling modes:
+ * 256x_Y_ -> 320x_Y_, H32/mode 4, PAR 5:4, for PAL DAR 4:3 (NTSC 7% aspect err)
+ * 256x224 -> 320x240, H32/mode 4, PAR 5:4, for NTSC DAR 4:3 (PAL 7% aspect err)
+ * 320x224 -> 320x240, PAR 1:1, for NTSC, DAR 4:3 (PAL 7% etc etc...)
+ * 160x144 -> 320x240: GG, PAR 6:5, scaling to 320x240 for DAR 4:3
*
* (C) 2021 kub <derkub@gmail.com>
*/
#include "upscale.h"
-/* 256x___ -> 320x___, H32/mode 4, PAR 5:4, for PAL DAR 4:3 (wrong for NTSC) */
-void upscale_clut_nn_256_320x___(u8 *__restrict di, int ds, u8 *__restrict si, int ss, int height)
+/* X x Y -> X*5/4 x Y */
+void upscale_clut_nn_x_4_5(u8 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height)
{
int y;
for (y = 0; y < height; y++) {
- h_upscale_nn_4_5(di, ds, si, ss, 256, f_nop);
+ h_upscale_nn_4_5(di, ds, si, ss, width, f_nop);
}
}
-void upscale_rgb_nn_256_320x___(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int height, u16 *pal)
+void upscale_rgb_nn_x_4_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
{
int y;
for (y = 0; y < height; y++) {
- h_upscale_nn_4_5(di, ds, si, ss, 256, f_pal);
+ h_upscale_nn_4_5(di, ds, si, ss, width, f_pal);
}
}
-void upscale_rgb_snn_256_320x___(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int height, u16 *pal)
+void upscale_rgb_snn_x_4_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
{
int y;
for (y = 0; y < height; y++) {
- h_upscale_snn_4_5(di, ds, si, ss, 256, f_pal);
+ h_upscale_snn_4_5(di, ds, si, ss, width, f_pal);
}
}
-void upscale_rgb_bl2_256_320x___(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int height, u16 *pal)
+void upscale_rgb_bl2_x_4_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
{
int y;
for (y = 0; y < height; y++) {
- h_upscale_bl2_4_5(di, ds, si, ss, 256, f_pal);
+ h_upscale_bl2_4_5(di, ds, si, ss, width, f_pal);
}
}
-void upscale_rgb_bl4_256_320x___(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int height, u16 *pal)
+void upscale_rgb_bl4_x_4_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
{
int y;
for (y = 0; y < height; y++) {
- h_upscale_bl4_4_5(di, ds, si, ss, 256, f_pal);
+ h_upscale_bl4_4_5(di, ds, si, ss, width, f_pal);
}
}
-/* 256x224 -> 320x240, H32/mode 4, PAR 5:4, for NTSC DAR 4:3 (wrong for PAL) */
-void upscale_clut_nn_256_320x224_240(u8 *__restrict di, int ds, u8 *__restrict si, int ss)
+/* X x Y -> X*5/4 x Y*17/16 */
+void upscale_clut_nn_x_4_5_y_16_17(u8 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height)
{
+ int swidth = width * 5/4;
int y, j;
- /* 14:15, 0 1 2 3 4 5 6 6 7 8 9 10 11 12 13 */
- for (y = 0; y < 224; y += 14) {
- /* lines 0-6 */
- for (j = 0; j < 7; j++) {
- h_upscale_nn_4_5(di, ds, si, ss, 256, f_nop);
+ for (y = 0; y < height; y += 16) {
+ for (j = 0; j < 8; j++) {
+ h_upscale_nn_4_5(di, ds, si, ss, width, f_nop);
}
- /* lines 8-14 */
- di += ds;
- for (j = 0; j < 7; j++) {
- h_upscale_nn_4_5(di, ds, si, ss, 256, f_nop);
+ di += ds;
+ for (j = 0; j < 8; j++) {
+ h_upscale_nn_4_5(di, ds, si, ss, width, f_nop);
}
- /* line 7 */
- di -= 8*ds;
- v_copy(&di[0], &di[-ds], 320, f_nop);
- di += 8*ds;
+
+ di -= 9*ds;
+ v_copy(&di[0], &di[-ds], swidth, f_nop);
+ di += 9*ds;
}
}
-void upscale_rgb_nn_256_320x224_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal)
+void upscale_rgb_nn_x_4_5_y_16_17(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
{
+ int swidth = width * 5/4;
int y, j;
- for (y = 0; y < 224; y += 14) {
- for (j = 0; j < 7; j++) {
- h_upscale_nn_4_5(di, ds, si, ss, 256, f_pal);
+ for (y = 0; y < height; y += 16) {
+ for (j = 0; j < 8; j++) {
+ h_upscale_nn_4_5(di, ds, si, ss, width, f_pal);
}
di += ds;
- for (j = 0; j < 7; j++) {
- h_upscale_nn_4_5(di, ds, si, ss, 256, f_pal);
+ for (j = 0; j < 8; j++) {
+ h_upscale_nn_4_5(di, ds, si, ss, width, f_pal);
}
- di -= 8*ds;
- v_copy(&di[0], &di[-ds], 320, f_nop);
- di += 8*ds;
+ di -= 9*ds;
+ v_copy(&di[0], &di[-ds], swidth, f_nop);
+ di += 9*ds;
}
}
-void upscale_rgb_snn_256_320x224_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal)
+void upscale_rgb_snn_x_4_5_y_16_17(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
{
+ int swidth = width * 5/4;
int y, j;
- /* 14:15, 0 1 2 3 4 5 5+6 6+7 7+8 8 9 10 11 12 13 */
- for (y = 0; y < 224; y += 14) {
- for (j = 0; j < 7; j++) {
- h_upscale_snn_4_5(di, ds, si, ss, 256, f_pal);
+ for (y = 0; y < height; y += 16) {
+ for (j = 0; j < 8; j++) {
+ h_upscale_snn_4_5(di, ds, si, ss, width, f_pal);
}
di += ds;
- for (j = 0; j < 7; j++) {
- h_upscale_snn_4_5(di, ds, si, ss, 256, f_pal);
+ for (j = 0; j < 8; j++) {
+ h_upscale_snn_4_5(di, ds, si, ss, width, f_pal);
}
/* mix lines 6-8 */
- di -= 8*ds;
- v_mix(&di[0], &di[-ds], &di[ds], 320, p_05, f_nop);
- v_mix(&di[-ds], &di[-2*ds], &di[-ds], 320, p_05, f_nop);
- v_mix(&di[ ds], &di[ ds], &di[ 2*ds], 320, p_05, f_nop);
- di += 8*ds;
+ di -= 9*ds;
+ v_mix(&di[0], &di[-ds], &di[ds], swidth, p_05, f_nop);
+ v_mix(&di[-ds], &di[-2*ds], &di[-ds], swidth, p_05, f_nop);
+ v_mix(&di[ ds], &di[ ds], &di[ 2*ds], swidth, p_05, f_nop);
+ di += 9*ds;
}
}
-void upscale_rgb_bln_256_320x224_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal)
+void upscale_rgb_bl2_x_4_5_y_16_17(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
{
+ int swidth = width * 5/4;
int y, j;
- /* 14:15, 0 1 2 2+3 3+4 4+5 5+6 6+7 7+8 8+9 9+10 10+11 11 12 13 */
- for (y = 0; y < 224; y += 14) {
- /* lines 0-2 */
- for (j = 0; j < 3; j++) {
- h_upscale_bln_4_5(di, ds, si, ss, 256, f_pal);
- }
- /* lines 3-11 mixing prep */
- di += ds;
- for (j = 0; j < 11; j++) {
- h_upscale_bln_4_5(di, ds, si, ss, 256, f_pal);
+ for (y = 0; y < height; y += 16) {
+ for (j = 0; j < 4; j++) {
+ h_upscale_bl2_4_5(di, ds, si, ss, width, f_pal);
}
- di -= 12*ds;
- /* mixing line 3: line 2 = -ds, line 3 = +ds */
- v_mix(&di[0], &di[-ds], &di[ds], 320, p_025, f_nop);
- di += ds;
- /* mixing lines 4-5: line n-1 = 0, line n = +ds */
- for (j = 0; j < 2; j++) {
- v_mix(&di[0], &di[0], &di[ds], 320, p_025, f_nop);
- di += ds;
- }
- /* mixing line 6-8 */
- for (j = 0; j < 3; j++) {
- v_mix(&di[0], &di[0], &di[ds], 320, p_05, f_nop);
- di += ds;
+ di += ds;
+ for (j = 0; j < 12; j++) {
+ h_upscale_bl2_4_5(di, ds, si, ss, width, f_pal);
}
- /* mixing lines 9-11 */
- for (j = 0; j < 3; j++) {
- v_mix(&di[0], &di[0], &di[ds], 320, p_075, f_nop);
+ /* mix lines 3-10 */
+ di -= 13*ds;
+ v_mix(&di[0], &di[-ds], &di[ds], swidth, p_05, f_nop);
+ for (j = 0; j < 7; j++) {
di += ds;
+ v_mix(&di[0], &di[0], &di[ds], swidth, p_05, f_nop);
}
- /* lines 12-14, already in place */
- di += 3*ds;
+ di += 6*ds;
}
}
-void upscale_rgb_bl2_256_320x224_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal)
+void upscale_rgb_bl4_x_4_5_y_16_17(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
{
+ int swidth = width * 5/4;
int y, j;
- /* 14:15, 0 1 2 2+3 3+4 4+5 5+6 6+7 7+8 8+9 9+10 10 11 12 13 */
- for (y = 0; y < 224; y += 14) {
- for (j = 0; j < 3; j++) {
- h_upscale_bl2_4_5(di, ds, si, ss, 256, f_pal);
+ for (y = 0; y < height; y += 16) {
+ for (j = 0; j < 2; j++) {
+ h_upscale_bl4_4_5(di, ds, si, ss, width, f_pal);
}
- di += ds;
- for (j = 0; j < 11; j++) {
- h_upscale_bl2_4_5(di, ds, si, ss, 256, f_pal);
+ di += ds;
+ for (j = 0; j < 14; j++) {
+ h_upscale_bl4_4_5(di, ds, si, ss, width, f_pal);
}
- /* mix lines 3-10 */
- di -= 12*ds;
- v_mix(&di[0], &di[-ds], &di[ds], 320, p_05, f_nop);
- for (j = 0; j < 7; j++) {
+ di -= 15*ds;
+ /* mixing line 2: line 1 = -ds, line 2 = +ds */
+ v_mix(&di[0], &di[-ds], &di[ds], swidth, p_025, f_nop);
+ di += ds;
+ /* mixing lines 3-5: line n-1 = 0, line n = +ds */
+ for (j = 0; j < 3; j++) {
+ v_mix(&di[0], &di[0], &di[ds], swidth, p_025, f_nop);
+ di += ds;
+ }
+ /* mixing lines 6-9 */
+ for (j = 0; j < 4; j++) {
+ v_mix(&di[0], &di[0], &di[ds], swidth, p_05, f_nop);
+ di += ds;
+ }
+ /* mixing lines 10-13 */
+ for (j = 0; j < 4; j++) {
+ v_mix(&di[0], &di[0], &di[ds], swidth, p_075, f_nop);
di += ds;
- v_mix(&di[0], &di[0], &di[ds], 320, p_05, f_nop);
}
- di += 5*ds;
+ /* lines 14-16, already in place */
+ di += 3*ds;
}
}
-void upscale_rgb_bl4_256_320x224_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal)
+/* "classic" upscaler as found in several emulators. It's really more like a
+ * x*4/3, y*16/15 upscaler, with an additional 5th row/17th line just inserted
+ * from the source image. That gives nice n/4,n/16 alpha values plus better
+ * symmetry in each block and avoids "borrowing" a row/line between blocks.
+ */
+void upscale_rgb_bln_x_4_5_y_16_17(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
{
+ int swidth = width * 5/4;
int y, j;
- /* 14:15, 0 0+1 1+2 2+3 3+4 4+5 5+6 6+7 7+8 8+9 9+10 10+11 11+12 12 13 */
- for (y = 0; y < 224; y += 14) {
- /* line 0 */
- h_upscale_bl4_4_5(di, ds, si, ss, 256, f_pal);
- /* lines 1-14 mixing prep */
+ for (y = 0; y < height; y += 16) {
+ for (j = 0; j < 4; j++) {
+ h_upscale_bln_4_5(di, ds, si, ss, width, f_pal);
+ }
di += ds;
- for (j = 0; j < 13; j++) {
- h_upscale_bl4_4_5(di, ds, si, ss, 256, f_pal);
+ for (j = 0; j < 12; j++) {
+ h_upscale_bln_4_5(di, ds, si, ss, width, f_pal);
}
- di -= 14*ds;
- /* mixing line 1: line 0 = -ds, line 1 = +ds */
- v_mix(&di[0], &di[-ds], &di[ds], 320, p_025, f_nop);
+ di -= 13*ds;
+ /* mixing line 4: line 3 = -ds, line 4 = +ds */
+ v_mix(&di[0], &di[-ds], &di[ds], swidth, p_025, f_nop);
di += ds;
- /* mixing lines 2-4: line n-1 = 0, line n = +ds */
- for (j = 0; j < 3; j++) {
- v_mix(&di[0], &di[0], &di[ds], 320, p_025, f_nop);
+ /* mixing lines 5-6: line n-1 = 0, line n = +ds */
+ for (j = 0; j < 2; j++) {
+ v_mix(&di[0], &di[0], &di[ds], swidth, p_025, f_nop);
di += ds;
}
- /* mixing lines 5-8 */
- for (j = 0; j < 4; j++) {
- v_mix(&di[0], &di[0], &di[ds], 320, p_05, f_nop);
+ /* mixing line 7-9 */
+ for (j = 0; j < 3; j++) {
+ v_mix(&di[0], &di[0], &di[ds], swidth, p_05, f_nop);
di += ds;
}
- /* mixing lines 9-12 */
- for (j = 0; j < 4; j++) {
- v_mix(&di[0], &di[0], &di[ds], 320, p_075, f_nop);
+ /* mixing lines 10-12 */
+ for (j = 0; j < 3; j++) {
+ v_mix(&di[0], &di[0], &di[ds], swidth, p_075, f_nop);
di += ds;
}
- /* lines 13-14, already in place */
- di += 2*ds;
+ /* lines 13-16, already in place */
+ di += 4*ds;
}
}
-void upscale_rgb_bl8_256_320x224_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal)
+/* experimental 8 level bilinear for quality assessment */
+void upscale_rgb_bl8_x_4_5_y_16_17(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
{
- int y, j, d;
+ int swidth = width * 5/4;
+ int y, j;
- /* 14:15, -1+0 0+1 1+2 2+3 3+4 4+5 5+6 6+7 7+8 8+9 9+10 10+11 11+12 12+13 13 */
- for (y = 0, d = ds; y < 224; y += 14, d = -ds) {
- /* lines 0-14 mixing prep */
+ for (y = 0; y < 224; y += 16) {
+ for (j = 0; j < 2; j++) {
+ h_upscale_bl8_4_5(di, ds, si, ss, width, f_pal);
+ }
di += ds;
for (j = 0; j < 14; j++) {
- h_upscale_bl8_4_5(di, ds, si, ss, 256, f_pal);
+ h_upscale_bl8_4_5(di, ds, si, ss, width, f_pal);
}
di -= 15*ds;
- /* mixing line 0: line 0 = -ds, line 1 = +ds */
- v_mix(&di[0], &di[d], &di[ds], 320, p_0125, f_nop);
+ /* mixing line 2: line 2 = -ds, line 3 = +ds */
+ v_mix(&di[0], &di[-ds], &di[ds], swidth, p_0125, f_nop);
di += ds;
- /* mixing line 1: line 1 = 0, line 2 = +ds */
- v_mix(&di[0], &di[0], &di[ds], 320, p_0125, f_nop);
+ /* mixing line 3: line 3 = 0, line 4 = +ds */
+ v_mix(&di[0], &di[0], &di[ds], swidth, p_0125, f_nop);
di += ds;
- /* mixing lines 2-3: line n-1 = 0, line n = +ds */
+ /* mixing lines 4-5: line n-1 = 0, line n = +ds */
for (j = 0; j < 2; j++) {
- v_mix(&di[0], &di[0], &di[ds], 320, p_025, f_nop);
+ v_mix(&di[0], &di[0], &di[ds], swidth, p_025, f_nop);
di += ds;
}
- /* mixing lines 4-5 */
+ /* mixing lines 6-7 */
for (j = 0; j < 2; j++) {
v_mix(&di[0], &di[0], &di[ds], 320, p_0375, f_nop);
di += ds;
}
- /* mixing lines 6-7 */
+ /* mixing lines 8-9 */
for (j = 0; j < 2; j++) {
v_mix(&di[0], &di[0], &di[ds], 320, p_05, f_nop);
di += ds;
}
- /* mixing lines 8-9 */
+ /* mixing lines 10-11 */
for (j = 0; j < 2; j++) {
v_mix(&di[0], &di[0], &di[ds], 320, p_0625, f_nop);
di += ds;
}
- /* mixing lines 10-11 */
+ /* mixing lines 12-13 */
for (j = 0; j < 2; j++) {
v_mix(&di[0], &di[0], &di[ds], 320, p_075, f_nop);
di += ds;
}
- /* mixing lines 12-13 */
+ /* mixing lines 14-15 */
for (j = 0; j < 2; j++) {
v_mix(&di[0], &di[0], &di[ds], 320, p_0875, f_nop);
di += ds;
}
- /* line 14, already in place */
+ /* line 16, already in place */
di += ds;
}
}
-/* 320x224 -> 320x240, PAR 1:1, for NTSC, DAR 4:3 (wrong for PAL) */
-void upscale_clut_nn_320x224_240(u8 *__restrict di, int ds, u8 *__restrict si, int ss)
+/* X x Y -> X x Y*17/16 */
+void upscale_clut_nn_y_16_17(u8 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height)
{
int y, j;
- for (y = 0; y < 224; y += 14) {
- for (j = 0; j < 7; j++) {
- h_copy(di, ds, si, ss, 320, f_nop);
+ for (y = 0; y < height; y += 16) {
+ for (j = 0; j < 8; j++) {
+ h_copy(di, ds, si, ss, width, f_nop);
}
- di += ds;
- for (j = 0; j < 7; j++) {
- h_copy(di, ds, si, ss, 320, f_nop);
+ di += ds;
+ for (j = 0; j < 8; j++) {
+ h_copy(di, ds, si, ss, width, f_nop);
}
- di -= 8*ds;
- v_copy(&di[0], &di[-ds], 320, f_nop);
- di += 8*ds;
-
+ di -= 9*ds;
+ v_copy(&di[0], &di[-ds], width, f_nop);
+ di += 9*ds;
}
}
-void upscale_rgb_nn_320x224_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal)
+void upscale_rgb_nn_y_16_17(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
{
int y, j;
- for (y = 0; y < 224; y += 14) {
- for (j = 0; j < 7; j++) {
- h_copy(di, ds, si, ss, 320, f_pal);
+ for (y = 0; y < height; y += 16) {
+ for (j = 0; j < 8; j++) {
+ h_copy(di, ds, si, ss, width, f_pal);
}
di += ds;
- for (j = 0; j < 7; j++) {
- h_copy(di, ds, si, ss, 320, f_pal);
+ for (j = 0; j < 8; j++) {
+ h_copy(di, ds, si, ss, width, f_pal);
}
- di -= 8*ds;
- v_copy(&di[0], &di[-ds], 320, f_nop);
- di += 8*ds;
+ di -= 9*ds;
+ v_copy(&di[0], &di[-ds], width, f_nop);
+ di += 9*ds;
}
}
-void upscale_rgb_snn_320x224_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal)
+void upscale_rgb_snn_y_16_17(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
{
int y, j;
- for (y = 0; y < 224; y += 14) {
- for (j = 0; j < 7; j++) {
- h_copy(di, ds, si, ss, 320, f_pal);
+ for (y = 0; y < height; y += 16) {
+ for (j = 0; j < 8; j++) {
+ h_copy(di, ds, si, ss, width, f_pal);
}
di += ds;
- for (j = 0; j < 7; j++) {
- h_copy(di, ds, si, ss, 320, f_pal);
+ for (j = 0; j < 8; j++) {
+ h_copy(di, ds, si, ss, width, f_pal);
}
- di -= 8*ds;
- v_mix(&di[ 0], &di[-ds], &di[ds], 320, p_05, f_nop);
- v_mix(&di[-ds], &di[-2*ds], &di[-ds], 320, p_05, f_nop);
- v_mix(&di[ ds], &di[ ds], &di[ 2*ds], 320, p_05, f_nop);
- di += 8*ds;
+ /* mix lines 6-8 */
+ di -= 9*ds;
+ v_mix(&di[0], &di[-ds], &di[ds], width, p_05, f_nop);
+ v_mix(&di[-ds], &di[-2*ds], &di[-ds], width, p_05, f_nop);
+ v_mix(&di[ ds], &di[ ds], &di[ 2*ds], width, p_05, f_nop);
+ di += 9*ds;
}
}
-void upscale_rgb_bl2_320x224_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal)
+void upscale_rgb_bl2_y_16_17(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
{
int y, j;
- for (y = 0; y < 224; y += 14) {
- for (j = 0; j < 3; j++) {
- h_copy(di, ds, si, ss, 320, f_pal);
+ for (y = 0; y < height; y += 16) {
+ for (j = 0; j < 4; j++) {
+ h_copy(di, ds, si, ss, width, f_pal);
}
- for (j = 0; j < 8; j++) {
- v_mix(&di[0], &si[-ss], &si[0], 320, p_05, f_pal);
- di += ds;
- si += ss;
+ di += ds;
+ for (j = 0; j < 12; j++) {
+ h_copy(di, ds, si, ss, width, f_pal);
}
- si -= ss;
- for (j = 0; j < 4; j++) {
- h_copy(di, ds, si, ss, 320, f_pal);
+ /* mix lines 3-10 */
+ di -= 13*ds;
+ v_mix(&di[0], &di[-ds], &di[ds], width, p_05, f_nop);
+ for (j = 0; j < 7; j++) {
+ di += ds;
+ v_mix(&di[0], &di[0], &di[ds], width, p_05, f_nop);
}
+ di += 6*ds;
}
}
-void upscale_rgb_bl4_320x224_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal)
+void upscale_rgb_bl4_y_16_17(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
{
int y, j;
- for (y = 0; y < 224; y += 14) {
- h_copy(di, ds, si, ss, 320, f_pal);
+ for (y = 0; y < height; y += 16) {
+ for (j = 0; j < 2; j++) {
+ h_copy(di, ds, si, ss, width, f_pal);
+ }
+ di += ds;
+ for (j = 0; j < 14; j++) {
+ h_copy(di, ds, si, ss, width, f_pal);
+ }
+ di -= 15*ds;
+ /* mixing line 2: line 1 = -ds, line 2 = +ds */
+ v_mix(&di[0], &di[-ds], &di[ds], width, p_025, f_nop);
+ di += ds;
+ /* mixing lines 3-5: line n-1 = 0, line n = +ds */
+ for (j = 0; j < 3; j++) {
+ v_mix(&di[0], &di[0], &di[ds], width, p_025, f_nop);
+ di += ds;
+ }
+ /* mixing lines 6-9 */
for (j = 0; j < 4; j++) {
- v_mix(&di[0], &si[-ss], &si[0], 320, p_025, f_pal);
+ v_mix(&di[0], &di[0], &di[ds], width, p_05, f_nop);
di += ds;
- si += ss;
}
+ /* mixing lines 10-13 */
for (j = 0; j < 4; j++) {
- v_mix(&di[0], &si[-ss], &si[0], 320, p_05, f_pal);
+ v_mix(&di[0], &di[0], &di[ds], width, p_075, f_nop);
di += ds;
- si += ss;
}
- for (j = 0; j < 4; j++) {
- v_mix(&di[0], &si[-ss], &si[0], 320, p_075, f_pal);
+ /* lines 14-16, already in place */
+ di += 3*ds;
+ }
+}
+
+/* X x Y -> X*2/1 x Y, e.g. for X 160->320 (GG) */
+void upscale_clut_nn_x_1_2(u8 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height)
+{
+ int y;
+
+ for (y = 0; y < height; y++) {
+ h_upscale_nn_1_2(di, ds, si, ss, width, f_nop);
+ }
+}
+
+void upscale_rgb_nn_x_1_2(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
+{
+ int y;
+
+ for (y = 0; y < height; y++) {
+ h_upscale_nn_1_2(di, ds, si, ss, width, f_pal);
+ }
+}
+
+void upscale_rgb_bl2_x_1_2(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
+{
+ int y;
+
+ for (y = 0; y < height; y++) {
+ h_upscale_bl2_1_2(di, ds, si, ss, width, f_pal);
+ }
+}
+
+/* X x Y -> X*2/1 x Y*5/3 (GG) */
+void upscale_clut_nn_x_1_2_y_3_5(u8 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height)
+{
+ int swidth = width * 2;
+ int y, j;
+
+ for (y = 0; y < height; y += 3) {
+ /* lines 0,2,4 */
+ for (j = 0; j < 3; j++) {
+ h_upscale_nn_1_2(di, ds, si, ss, width, f_nop);
di += ds;
- si += ss;
}
- si -= ss;
+ /* lines 1,3 */
+ di -= 5*ds;
for (j = 0; j < 2; j++) {
- h_copy(di, ds, si, ss, 320, f_pal);
+ v_copy(&di[0], &di[-ds], swidth, f_nop);
+ di += 2*ds;
}
}
}
-/* 160x144 -> 320x240: GG, PAR 6:5, scaling to 320x240 for DAR 4:3 */
-/* NB for smoother image could scale to 288x216, x*9/5, y*3/2 ?
- * h: 11111 11112 22222 22233 33333 33444 44444 45555 55555
- * 1 1 2 2+3 3 3+4 4 5 5
- * v: 11 12 22
- * 1 1+2 2
- */
-void upscale_clut_nn_160_320x144_240(u8 *__restrict di, int ds, u8 *__restrict si, int ss)
+void upscale_rgb_nn_x_1_2_y_3_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
{
+ int swidth = width * 2;
int y, j;
- /* 3:5, 0 0 1 1 2 */
- for (y = 0; y < 144; y += 3) {
- /* lines 0,2,4 */
+ for (y = 0; y < height; y += 3) {
for (j = 0; j < 3; j++) {
- h_upscale_nn_1_2(di, ds, si, ss, 160, f_nop);
+ h_upscale_nn_1_2(di, ds, si, ss, width, f_pal);
+ di += ds;
+ }
+ di -= 5*ds;
+ for (j = 0; j < 2; j++) {
+ v_copy(&di[0], &di[-ds], swidth, f_nop);
+ di += 2*ds;
+ }
+ }
+}
+
+void upscale_rgb_bl2_x_1_2_y_3_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
+{
+ int swidth = width * 2;
+ int y, j;
+
+ for (y = 0; y < height; y += 3) {
+ for (j = 0; j < 3; j++) {
+ h_upscale_bl2_1_2(di, ds, si, ss, width, f_pal);
di += ds;
}
- /* lines 1,3 */
di -= 5*ds;
for (j = 0; j < 2; j++) {
- v_copy(&di[0], &di[-ds], 320, f_nop);
+ v_mix(&di[0], &di[-ds], &di[ds], swidth, p_05, f_nop);
di += 2*ds;
}
}
}
-void upscale_rgb_nn_160_320x144_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal)
+void upscale_rgb_bl4_x_1_2_y_3_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
+{
+ int swidth = width * 2;
+ int y, j, d;
+
+ /* for 1st block backwards reference virtually duplicate source line 0 */
+ for (y = 0, d = 2*ds; y < height; y += 3, d = -ds) {
+ di += 2*ds;
+ for (j = 0; j < 3; j++) {
+ h_upscale_bl2_1_2(di, ds, si, ss, width, f_pal);
+ }
+ di -= 5*ds;
+ v_mix(&di[0], &di[d ], &di[2*ds], swidth, p_05, f_nop); /*-1+0 */
+ di += ds;
+ v_mix(&di[0], &di[ds], &di[2*ds], swidth, p_075, f_nop);/* 0+1 */
+ di += ds;
+ v_mix(&di[0], &di[ 0], &di[ ds], swidth, p_025, f_nop);/* 0+1 */
+ di += ds;
+ v_mix(&di[0], &di[ 0], &di[ ds], swidth, p_05, f_nop); /* 1+2 */
+ di += 2*ds;
+ }
+}
+
+/* X x Y -> X x Y*5/3, e.g. for Y 144->240 (GG) */
+void upscale_clut_nn_y_3_5(u8 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height)
{
int y, j;
- for (y = 0; y < 144; y += 3) {
+ for (y = 0; y < height; y += 3) {
+ /* lines 0,2,4 */
for (j = 0; j < 3; j++) {
- h_upscale_nn_1_2(di, ds, si, ss, 160, f_pal);
+ h_copy(di, ds, si, ss, width, f_nop);
di += ds;
}
+ /* lines 1,3 */
di -= 5*ds;
for (j = 0; j < 2; j++) {
- v_copy(&di[0], &di[-ds], 320, f_nop);
+ v_copy(&di[0], &di[-ds], width, f_nop);
di += 2*ds;
}
}
}
-void upscale_rgb_snn_160_320x144_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal)
+void upscale_rgb_nn_y_3_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
{
int y, j;
- /* 3:5, 0 0+1 1 1+2 2 */
- for (y = 0; y < 144; y += 3) {
+ for (y = 0; y < height; y += 3) {
for (j = 0; j < 3; j++) {
- h_upscale_nn_1_2(di, ds, si, ss, 160, f_pal);
+ h_copy(di, ds, si, ss, width, f_pal);
di += ds;
}
di -= 5*ds;
for (j = 0; j < 2; j++) {
- v_mix(&di[0], &di[-ds], &di[ds], 320, p_05, f_nop);
+ v_copy(&di[0], &di[-ds], width, f_nop);
di += 2*ds;
}
}
}
-void upscale_rgb_bl2_160_320x144_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal)
+void upscale_rgb_bl2_y_3_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
{
int y, j;
- /* 3:5, 0 0+1 1 1+2 2 */
- for (y = 0; y < 144; y += 3) {
+ for (y = 0; y < height; y += 3) {
for (j = 0; j < 3; j++) {
- h_upscale_bl2_1_2(di, ds, si, ss, 160, f_pal);
+ h_copy(di, ds, si, ss, width, f_pal);
di += ds;
}
di -= 5*ds;
for (j = 0; j < 2; j++) {
- v_mix(&di[0], &di[-ds], &di[ds], 320, p_05, f_nop);
+ v_mix(&di[0], &di[-ds], &di[ds], width, p_05, f_nop);
di += 2*ds;
}
}
}
-void upscale_rgb_bl4_160_320x144_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal)
+void upscale_rgb_bl4_y_3_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal)
{
int y, j, d;
- /* 3:5, -1+0, 0+1 0+1 1+2 2
- * for 1st block backwards reference virtually duplicate source line 0 */
- for (y = 0, d = 2*ds; y < 144; y += 3, d = -ds) {
+ /* for 1st block backwards reference virtually duplicate source line 0 */
+ for (y = 0, d = 2*ds; y < height; y += 3, d = -ds) {
di += 2*ds;
for (j = 0; j < 3; j++) {
- h_upscale_bl2_1_2(di, ds, si, ss, 160, f_pal);
+ h_copy(di, ds, si, ss, width, f_pal);
}
di -= 5*ds;
- v_mix(&di[0], &di[d ], &di[2*ds], 320, p_05, f_nop); /*-1+0 */
+ v_mix(&di[0], &di[d ], &di[2*ds], width, p_05, f_nop); /*-1+0 */
di += ds;
- v_mix(&di[0], &di[ds], &di[2*ds], 320, p_075, f_nop); /* 0+1 */
+ v_mix(&di[0], &di[ds], &di[2*ds], width, p_075, f_nop);/* 0+1 */
di += ds;
- v_mix(&di[0], &di[ 0], &di[ ds], 320, p_025, f_nop); /* 0+1 */
+ v_mix(&di[0], &di[ 0], &di[ ds], width, p_025, f_nop);/* 0+1 */
di += ds;
- v_mix(&di[0], &di[ 0], &di[ ds], 320, p_05, f_nop); /* 1+2 */
+ v_mix(&di[0], &di[ 0], &di[ ds], width, p_05, f_nop); /* 1+2 */
di += 2*ds;
}
}
* nn: nearest neighbour
* snn: "smoothed" nearest neighbour (see below)
* bln: n-level-bilinear with n quantized weights
- * quantization: 0: a<1/2*n, 1/n: 1/2*n<=a<3/2*n, etc
+ * quantization: 0: a<1/(2*n), 1/n: 1/(2*n)<=a<3/(2*n), etc
* currently n=2, n=4 are implemented (there's n=8 mixing, but no filters)
* [NB this has been brought to my attn, which is probably the same as bl2:
* https://www.drdobbs.com/image-scaling-with-bresenham/184405045?pgno=1]
* a sharper look than a bilinear filter, at the price of some visible jags
* on diagonal edges.
*
- * scaling modes:
- * 256x___ -> 320x___ only horizontal scaling. Produces an aspect error of
- * ~7% for NTSC 224 line modes, but is correct for PAL
- * 256/320x224/240
- * -> 320x240 always produces 320x240 at DAR 4:3
-* 160x144 -> 320x240 game gear (currently unused)
-*
+ * example scaling modes:
+ * 256x_Y_ -> 320x_Y_, H32/mode 4, PAR 5:4, for PAL DAR 4:3 (NTSC 7% aspect err)
+ * 256x224 -> 320x240, H32/mode 4, PAR 5:4, for NTSC DAR 4:3 (PAL 7% aspect err)
+ * 320x224 -> 320x240, PAR 1:1, for NTSC, DAR 4:3 (PAL 7% etc etc...)
+ * 160x144 -> 320x240: GG, PAR 6:5, scaling to 320x240 for DAR 4:3
+ *
+ *
* (C) 2021 kub <derkub@gmail.com>
+ *
+ * This work is licensed under the terms of MAME license.
+ * See COPYING file in the top-level directory.
*/
#include <pico/pico_types.h>
/* RGB565 pixel mixing, see https://www.compuphase.com/graphic/scale3.htm and
http://blargg.8bitalley.com/info/rgb_mixing.html */
/* 2-level mixing */
-//#define p_05(p1,p2) (((p1)+(p2) + ( ((p1)^(p2))&0x0821))>>1) // round up
-//#define p_05(p1,p2) (((p1)+(p2) - ( ((p1)^(p2))&0x0821))>>1) // round down
-#define p_05(p1,p2) (((p1)&(p2)) + ((((p1)^(p2))&~0x0821)>>1))
+//#define p_05(d,p1,p2) d=(((p1)+(p2) + ( ((p1)^(p2))&0x0821))>>1) // round up
+//#define p_05(d,p1,p2) d=(((p1)+(p2) - ( ((p1)^(p2))&0x0821))>>1) // round down
+#define p_05(d,p1,p2) d=(((p1)&(p2)) + ((((p1)^(p2))&~0x0821)>>1))
/* 4-level mixing, 2 times slower */
// 1/4*p1 + 3/4*p2 = 1/2*(1/2*(p1+p2) + p2)
-#define p_025(p1,p2) (t=p_05(p1, p2), p_05( t, p2))
-#define p_075(p1,p2) p_025(p2,p1)
+#define p_025(d,p1,p2) p_05(t, p1, p2); p_05( d, t, p2)
+#define p_075(d,p1,p2) p_025(d,p2,p1)
/* 8-level mixing, 3 times slower */
// 1/8*p1 + 7/8*p2 = 1/2*(1/2*(1/2*(p1+p2) + p2) + p2)
-#define p_0125(p1,p2) (t=p_05(p1, p2), u=p_05( t, p2), p_05( u, p2))
+#define p_0125(d,p1,p2) p_05(t, p1, p2); p_05( u, t, p2); p_05( d, u, p2)
// 3/8*p1 + 5/8*p2 = 1/2*(1/2*(1/2*(p1+p2) + p2) + 1/2*(p1+p2))
-#define p_0375(p1,p2) (t=p_05(p1, p2), u=p_05( t, p2), p_05( u, t))
-#define p_0625(p1,p2) p_0375(p2,p1)
-#define p_0875(p1,p2) p_0125(p2,p1)
+#define p_0375(d,p1,p2) p_05(t, p1, p2); p_05( u, t, p2); p_05( d, u, t)
+#define p_0625(d,p1,p2) p_0375(d,p2,p1)
+#define p_0875(d,p1,p2) p_0125(d,p2,p1)
/* pixel transforms */
#define f_pal(v) pal[v] // convert CLUT index -> RGB565
/*
scalers h:
256->320: - (4:5) (256x224/240 -> 320x224/240)
-256->299: - (6:7) (256x224 -> 299x224, DAR 4:3, 10.5 px border )
+256->299: - (6:7) (256x224 -> 299x224, alt?)
160->320: - (1:2) 2x (160x144 -> 320x240, GG)
160->288: - (5:9) (160x144 -> 288x216, GG alt?)
*/
for (i = w/4; i > 0; i--, si += 4, di += 5) { \
di[0] = f(si[0]); \
di[1] = f(si[1]); \
- di[2] = p_05(f(si[1]),f(si[2])); \
+ p_05(di[2], f(si[1]),f(si[2])); \
di[3] = f(si[2]); \
di[4] = f(si[3]); \
} \
} while (0)
#define h_upscale_bln_4_5(di,ds,si,ss,w,f) do { \
- int i, t; \
+ int i; u16 t; \
for (i = w/4; i > 0; i--, si += 4, di += 5) { \
di[0] = f(si[0]); \
- di[1] = p_025(f(si[0]),f(si[1])); \
- di[2] = p_05 (f(si[1]),f(si[2])); \
- di[3] = p_075(f(si[2]),f(si[3])); \
+ p_025(di[1], f(si[0]),f(si[1])); \
+ p_05 (di[2], f(si[1]),f(si[2])); \
+ p_075(di[3], f(si[2]),f(si[3])); \
di[4] = f(si[3]); \
} \
di += ds - w/4*5; \
int i; \
for (i = w/4; i > 0; i--, si += 4, di += 5) { \
di[0] = f(si[0]); \
- di[1] = p_05(f(si[0]),f(si[1])); \
- di[2] = p_05(f(si[1]),f(si[2])); \
+ p_05(di[1], f(si[0]),f(si[1])); \
+ p_05(di[2], f(si[1]),f(si[2])); \
di[3] = f(si[2]); \
di[4] = f(si[3]); \
} \
} while (0)
#define h_upscale_bl4_4_5(di,ds,si,ss,w,f) do { \
- int i, t; uint p = f(si[0]); \
+ int i; u16 t, p = f(si[0]); \
for (i = w/4; i > 0; i--, si += 4, di += 5) { \
- di[0] = p_025(p, f(si[0])); \
- di[1] = p_05 (f(si[0]),f(si[1])); \
- di[2] = p_05 (f(si[1]),f(si[2])); \
- di[3] = p_075(f(si[2]),f(si[3])); \
+ p_025(di[0], p, f(si[0])); \
+ p_05 (di[1], f(si[0]),f(si[1])); \
+ p_05 (di[2], f(si[1]),f(si[2])); \
+ p_075(di[3], f(si[2]),f(si[3])); \
di[4] = p = f(si[3]); \
} \
di += ds - w/4*5; \
} while (0)
#define h_upscale_bl8_4_5(di,ds,si,ss,w,f) do { \
- int i, t, u; uint p = f(si[0]); \
+ int i; u16 t, u, p = f(si[0]); \
for (i = w/4; i > 0; i--, si += 4, di += 5) { \
- di[0] = p_025(p, f(si[0])); \
- di[1] = p_0375(f(si[0]),f(si[1])); \
- di[2] = p_0625(f(si[1]),f(si[2])); \
- di[3] = p_075(f(si[2]),f(si[3])); \
+ p_025 (di[0], p, f(si[0])); \
+ p_0375(di[1], f(si[0]),f(si[1])); \
+ p_0625(di[2], f(si[1]),f(si[2])); \
+ p_075 (di[3], f(si[2]),f(si[3])); \
di[4] = p = f(si[3]); \
} \
di += ds - w/4*5; \
di[0] = f(si[0]); \
di[1] = f(si[1]); \
di[2] = f(si[2]); \
- di[3] = p_05(f(si[2]),f(si[3])); \
+ p_05(di[3], f(si[2]),f(si[3])); \
di[4] = f(si[3]); \
di[5] = f(si[4]); \
di[6] = f(si[5]); \
int i; \
for (i = w/6; i > 0; i--, si += 6, di += 7) { \
di[0] = f(si[0]); \
- di[1] = p_05(f(si[0]),f(si[1])); \
- di[2] = p_05(f(si[1]),f(si[2])); \
- di[3] = p_05(f(si[2]),f(si[3])); \
- di[4] = p_05(f(si[3]),f(si[4])); \
+ p_05(di[1], f(si[0]),f(si[1])); \
+ p_05(di[2], f(si[1]),f(si[2])); \
+ p_05(di[3], f(si[2]),f(si[3])); \
+ p_05(di[4], f(si[3]),f(si[4])); \
di[5] = f(si[4]); \
di[6] = f(si[5]); \
} \
} while (0)
#define h_upscale_bl4_6_7(di,ds,si,ss,w,f) do { \
- int i, t; uint p = f(si[0]); \
+ int i; u16 t p = f(si[0]); \
for (i = w/6; i > 0; i--, si += 6, di += 7) { \
- di[0] = p_025(p,f(si[0])); \
- di[1] = p_025(f(si[0]),f(si[1])); \
- di[2] = p_05 (f(si[1]),f(si[2])); \
- di[3] = p_05 (f(si[2]),f(si[3])); \
- di[4] = p_075(f(si[3]),f(si[4])); \
- di[5] = p_075(f(si[4]),f(si[5])); \
+ p_025(di[0], p, f(si[0])); \
+ p_025(di[1], f(si[0]),f(si[1])); \
+ p_05 (di[2], f(si[1]),f(si[2])); \
+ p_05 (di[3], f(si[2]),f(si[3])); \
+ p_075(di[4], f(si[3]),f(si[4])); \
+ p_075(di[5], f(si[4]),f(si[5])); \
di[6] = p = f(si[5]); \
} \
di += ds - w/6*7; \
di[0] = f(si[0]); \
di[1] = f(si[0]); \
di[2] = f(si[1]); \
- di[3] = p_05(f(si[1]),f(si[2])); \
+ p_05(di[3], f(si[1]),f(si[2])); \
di[4] = f(si[2]); \
- di[5] = p_05(f(si[2]),f(si[3])); \
+ p_05(di[5], f(si[2]),f(si[3])); \
di[6] = f(si[3]); \
di[7] = f(si[4]); \
di[8] = f(si[4]); \
int i; \
for (i = w/5; i > 0; i--, si += 5, di += 9) { \
di[0] = f(si[0]); \
- di[1] = p_05(f(si[0]),f(si[1])); \
+ p_05(di[1], f(si[0]),f(si[1])); \
di[2] = f(si[1]); \
- di[3] = p_05(f(si[1]),f(si[2])); \
+ p_05(di[3], f(si[1]),f(si[2])); \
di[4] = f(si[2]); \
- di[5] = p_05(f(si[2]),f(si[3])); \
+ p_05(di[5], f(si[2]),f(si[3])); \
di[6] = f(si[3]); \
- di[7] = p_05(f(si[3]),f(si[4])); \
+ p_05(di[7], f(si[3]),f(si[4])); \
di[8] = f(si[4]); \
} \
di += ds - w/5*9; \
} while (0)
#define h_upscale_bl4_5_9(di,ds,si,ss,w,f) do { \
- int i, t; uint p = f(si[0]); \
+ int i; u16 t, p = f(si[0]); \
for (i = w/5; i > 0; i--, si += 5, di += 9) { \
- di[0] = p_05 (p,f(si[0])); \
+ p_05 (di[0], p, f(si[0])); \
di[1] = f(si[0]); \
- di[2] = p_025(f(si[0]),f(si[1])); \
- di[3] = p_075(f(si[1]),f(si[2])); \
- di[4] = p_025(f(si[1]),f(si[2])); \
- di[5] = p_075(f(si[2]),f(si[3])); \
+ p_025(di[2], f(si[0]),f(si[1])); \
+ p_075(di[3], f(si[1]),f(si[2])); \
+ p_025(di[4], f(si[1]),f(si[2])); \
+ p_075(di[5], f(si[2]),f(si[3])); \
di[6] = f(si[3]); \
- di[7] = p_05 (f(si[3]),f(si[4])); \
+ p_05 (di[7], f(si[3]),f(si[4])); \
di[8] = p = f(si[4]); \
} \
di += ds - w/5*9; \
#define h_upscale_bl2_1_2(di,ds,si,ss,w,f) do { \
int i; uint p = f(si[0]); \
for (i = w/2; i > 0; i--, si += 2, di += 4) { \
- di[0] = p_05 (p, f(si[0])); \
+ p_05 (di[0], p, f(si[0])); \
di[1] = f(si[0]); \
- di[2] = p_05 (f(si[0]), f(si[1])); \
+ p_05 (di[2], f(si[0]),f(si[1])); \
di[3] = p = f(si[1]); \
} \
di += ds - w*2; \
*/
#define v_mix(di,li,ri,w,p_mix,f) do { \
- int i, t, u; (void)t, (void)u; \
+ u16 i, t, u; (void)t, (void)u; \
for (i = 0; i < w; i += 4) { \
- (di)[i ] = p_mix(f((li)[i ]), f((ri)[i ])); \
- (di)[i+1] = p_mix(f((li)[i+1]), f((ri)[i+1])); \
- (di)[i+2] = p_mix(f((li)[i+2]), f((ri)[i+2])); \
- (di)[i+3] = p_mix(f((li)[i+3]), f((ri)[i+3])); \
+ p_mix((di)[i ], f((li)[i ]),f((ri)[i ])); \
+ p_mix((di)[i+1], f((li)[i+1]),f((ri)[i+1])); \
+ p_mix((di)[i+2], f((li)[i+2]),f((ri)[i+2])); \
+ p_mix((di)[i+3], f((li)[i+3]),f((ri)[i+3])); \
} \
} while (0)
} \
} while (0)
+/* scale 14:15 */
+#define v_upscale_nn_14_15(di,ds,w,l) do { \
+ if (++l == 7) { \
+ di += ds; \
+ } else if (l >= 14) { \
+ l = 0; \
+ di -= 7*ds; \
+ v_copy(&di[0], &di[-ds], w, f_nop); \
+ di += 7*ds; \
+ } \
+} while (0)
+
+#define v_upscale_snn_14_15(di,ds,w,l) do { \
+ if (++l == 7) { \
+ di += ds; \
+ } else if (l >= 14) { \
+ l = 0; \
+ di -= 7*ds; \
+ v_mix(&di[0], &di[-ds], &di[ds], w, p_05, f_nop); \
+ v_mix(&di[-ds], &di[-2*ds], &di[-ds], w, p_05, f_nop); \
+ v_mix(&di[ ds], &di[ ds], &di[ 2*ds], w, p_05, f_nop); \
+ di += 7*ds; \
+ } \
+} while (0)
+
+#define v_upscale_bl2_14_15(di,ds,w,l) do { \
+ if (++l == 3) { \
+ di += ds; \
+ } else if (l >= 14) { \
+ int j; \
+ l = 0; \
+ di -= 11*ds; \
+ v_mix(&di[0], &di[-ds], &di[ds], w, p_05, f_nop); \
+ for (j = 0; j < 7; j++) { \
+ di += ds; \
+ v_mix(&di[0], &di[0], &di[ds], w, p_05, f_nop); \
+ } \
+ di += 4*ds; \
+ } \
+} while (0)
+
+#define v_upscale_bl4_14_15(di,ds,w,l) do { \
+ if (++l == 1) { \
+ di += ds; \
+ } else if (l >= 14) { \
+ int j; \
+ l = 0; \
+ di -= 13*ds; \
+ v_mix(&di[0], &di[-ds], &di[ds], w, p_025, f_nop); \
+ di += ds; \
+ for (j = 0; j < 3; j++) { \
+ v_mix(&di[0], &di[0], &di[ds], w, p_025, f_nop); \
+ di += ds; \
+ } \
+ for (j = 0; j < 4; j++) { \
+ v_mix(&di[0], &di[0], &di[ds], w, p_05, f_nop); \
+ di += ds; \
+ } \
+ for (j = 0; j < 4; j++) { \
+ v_mix(&di[0], &di[0], &di[ds], w, p_075, f_nop); \
+ di += ds; \
+ } \
+ di += 1*ds; \
+ } \
+} while (0)
+
+/* scale 16:17 */
+#define v_upscale_nn_16_17(di,ds,w,l) do { \
+ if (++l == 8) { \
+ di += ds; \
+ } else if (l >= 16) { \
+ l = 0; \
+ di -= 8*ds; \
+ v_copy(&di[0], &di[-ds], w, f_nop); \
+ di += 8*ds; \
+ } \
+} while (0)
+
+#define v_upscale_snn_16_17(di,ds,w,l) do { \
+ if (++l == 8) { \
+ di += ds; \
+ } else if (l >= 16) { \
+ l = 0; \
+ di -= 8*ds; \
+ v_mix(&di[0], &di[-ds], &di[ds], w, p_05, f_nop); \
+ v_mix(&di[-ds], &di[-2*ds], &di[-ds], w, p_05, f_nop); \
+ v_mix(&di[ ds], &di[ ds], &di[ 2*ds], w, p_05, f_nop); \
+ di += 8*ds; \
+ } \
+} while (0)
+
+#define v_upscale_bl2_16_17(di,ds,w,l) do { \
+ if (++l == 4) { \
+ di += ds; \
+ } else if (l >= 16) { \
+ int j; \
+ l = 0; \
+ di -= 12*ds; \
+ v_mix(&di[0], &di[-ds], &di[ds], w, p_05, f_nop); \
+ for (j = 0; j < 7; j++) { \
+ di += ds; \
+ v_mix(&di[0], &di[0], &di[ds], w, p_05, f_nop); \
+ } \
+ di += 5*ds; \
+ } \
+} while (0)
+
+#define v_upscale_bl4_16_17(di,ds,w,l) do { \
+ if (++l == 2) { \
+ di += ds; \
+ } else if (l >= 16) { \
+ int j; \
+ l = 0; \
+ di -= 14*ds; \
+ v_mix(&di[0], &di[-ds], &di[ds], w, p_025, f_nop); \
+ di += ds; \
+ for (j = 0; j < 3; j++) { \
+ v_mix(&di[0], &di[0], &di[ds], w, p_025, f_nop); \
+ di += ds; \
+ } \
+ for (j = 0; j < 4; j++) { \
+ v_mix(&di[0], &di[0], &di[ds], w, p_05, f_nop); \
+ di += ds; \
+ } \
+ for (j = 0; j < 4; j++) { \
+ v_mix(&di[0], &di[0], &di[ds], w, p_075, f_nop); \
+ di += ds; \
+ } \
+ di += 2*ds; \
+ } \
+} while (0)
+
+/* scale 3:5 */
+#define v_upscale_nn_3_5(di,ds,w,l) do { \
+ if (++l < 3) { \
+ di += ds; \
+ } else { \
+ int j; \
+ l = 0; \
+ di -= 4*ds; \
+ for (j = 0; j < 2; j++) { \
+ v_copy(&di[0], &di[-ds], w, f_nop); \
+ di += 2*ds; \
+ } \
+ } \
+} while (0)
+
+#define v_upscale_snn_3_5(di,ds,w,l) do { \
+ if (++l < 3) { \
+ di += ds; \
+ } else { \
+ int j; \
+ l = 0; \
+ di -= 4*ds; \
+ for (j = 0; j < 2; j++) { \
+ v_mix(&di[0], &di[-ds], &di[ds], w, p_05, f_nop); \
+ di += 2*ds; \
+ } \
+ } \
+} while (0)
+
+/* scale 2:3 */
+#define v_upscale_nn_2_3(di,ds,w,l) do { \
+ if (++l < 2) { \
+ di += ds; \
+ } else { \
+ int j; \
+ l = 0; \
+ di -= 2*ds; \
+ v_copy(&di[0], &di[-ds], w, f_nop); \
+ di += 2*ds; \
+ } \
+} while (0)
+
+#define v_upscale_snn_2_3(di,ds,w,l) do { \
+ if (++l < 2) { \
+ di += ds; \
+ } else { \
+ int j; \
+ l = 0; \
+ di -= 2*ds; \
+ v_mix(&di[0], &di[-ds], &di[ds], w, p_05, f_nop); \
+ di += 2*ds; \
+ } \
+} while (0)
+
+
+/* X x Y -> X*5/4 x Y, for X 256->320 */
+void upscale_rgb_nn_x_4_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
+void upscale_rgb_snn_x_4_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
+void upscale_rgb_bl2_x_4_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
+void upscale_rgb_bl4_x_4_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
+
+/* X x Y -> X x Y*17/16, for Y 224->238 or 192->204 (SMS) */
+void upscale_rgb_nn_y_16_17(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
+void upscale_rgb_snn_y_16_17(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
+void upscale_rgb_bl2_y_16_17(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
+void upscale_rgb_bl4_y_16_17(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
+/* X x Y -> X*5/4 x Y*17/16 */
+void upscale_rgb_nn_x_4_5_y_16_17(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
+void upscale_rgb_snn_x_4_5_y_16_17(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
+void upscale_rgb_bl2_x_4_5_y_16_17(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
+void upscale_rgb_bl4_x_4_5_y_16_17(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
-/* 256x___ -> 320x___, H32/mode 4, PAR 5:4, for PAL DAR 4:3 (wrong for NTSC) */
-void upscale_clut_nn_256_320x___(u8 *__restrict di, int ds, u8 *__restrict si, int ss, int height);
-void upscale_rgb_nn_256_320x___(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int height, u16 *pal);
-void upscale_rgb_snn_256_320x___(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int height, u16 *pal);
-void upscale_rgb_bl2_256_320x___(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int height, u16 *pal);
-void upscale_rgb_bl4_256_320x___(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int height, u16 *pal);
+/* X x Y -> X*2/1 x Y, e.g. for X 160->320 (GG) */
+void upscale_rgb_nn_x_1_2(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
+void upscale_rgb_bl2_x_1_2(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
-/* 256x224 -> 320x240, H32/mode 4, PAR 5:4, for NTSC DAR 4:3 (wrong for PAL) */
-void upscale_clut_nn_256_320x224_240(u8 *__restrict di, int ds, u8 *__restrict si, int ss);
-void upscale_rgb_nn_256_320x224_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal);
-void upscale_rgb_snn_256_320x224_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal);
-void upscale_rgb_bl2_256_320x224_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal);
-void upscale_rgb_bl4_256_320x224_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal);
+/* X x Y -> X x Y*5/3, e.g. for Y 144->240 (GG) */
+void upscale_rgb_nn_y_3_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
+void upscale_rgb_bl2_y_3_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
+void upscale_rgb_bl4_y_3_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
-/* 320x224 -> 320x240, PAR 1:1, for NTSC, DAR 4:3 (wrong for PAL) */
-void upscale_clut_nn_320x224_240(u8 *__restrict di, int ds, u8 *__restrict si, int ss);
-void upscale_rgb_nn_320x224_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal);
-void upscale_rgb_snn_320x224_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal);
-void upscale_rgb_bl2_320x224_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal);
-void upscale_rgb_bl4_320x224_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal);
+/* X x Y -> X*2/1 x Y*5/3 (GG) */
+void upscale_rgb_nn_x_1_2_y_3_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
+void upscale_rgb_bl2_x_1_2_y_3_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
+void upscale_rgb_bl4_x_1_2_y_3_5(u16 *__restrict di, int ds, u8 *__restrict si, int ss, int width, int height, u16 *pal);
-/* 160x144 -> 320x240: GG, PAR 6:5, scaling to 320x240 for DAR 4:3 */
-void upscale_clut_nn_160_320x144_240(u8 *__restrict di, int ds, u8 *__restrict si, int ss);
-void upscale_rgb_nn_160_320x144_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal);
-void upscale_rgb_snn_160_320x144_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal);
-void upscale_rgb_bl2_160_320x144_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal);
-void upscale_rgb_bl4_160_320x144_240(u16 *__restrict di, int ds, u8 *__restrict si, int ss, u16 *pal);
Pico.m.dirtyPal = 1;\r
\r
PicoIn.opt &= ~POPT_EN_SOFTSCALE;\r
- if (currentConfig.scaling == EOPT_SCALE_SW)\r
+ if (currentConfig.scaling == EOPT_SCALE_SW) {\r
PicoIn.opt |= POPT_EN_SOFTSCALE;\r
+ PicoIn.filter = EOPT_FILTER_BILINEAR2;\r
+ }\r
\r
// palette converters for 8bit modes\r
make_local_pal = (PicoIn.AHW & PAHW_SMS) ? make_local_pal_sms : make_local_pal_md;\r
}\r
\r
-void emu_video_mode_change(int start_line, int line_count, int is_32cols)\r
+void emu_video_mode_change(int start_line, int line_count, int start_col, int col_count)\r
{\r
int scalex = 320, scaley = 240;\r
int ln_offs = 0;\r
\r
/* set up hwscaling here */\r
PicoIn.opt &= ~POPT_DIS_32C_BORDER;\r
- if (is_32cols && currentConfig.scaling == EOPT_SCALE_HW) {\r
- scalex = 256;\r
+ if (col_count < 320 && currentConfig.scaling == EOPT_SCALE_HW) {\r
+ scalex = col_count;\r
PicoIn.opt |= POPT_DIS_32C_BORDER;\r
- osd_fps_x = OSD_FPS_X - 64;\r
+ osd_fps_x = col_count - (320-OSD_FPS_X);\r
}\r
\r
if (currentConfig.vscaling == EOPT_SCALE_HW) {\r
static retro_audio_sample_batch_t audio_batch_cb;
#define VOUT_MAX_WIDTH 320
-#define VOUT_32COL_WIDTH 256
#define VOUT_MAX_HEIGHT 240
#define INITIAL_SND_RATE 44100
/* Required to allow on the fly changes to 'show overscan' */
static int vm_current_start_line = -1;
static int vm_current_line_count = -1;
-static int vm_current_is_32cols = -1;
+static int vm_current_start_col = -1;
+static int vm_current_col_count = -1;
static int vout_16bit = 1;
static int vout_format = PDF_RGB555;
return ret;
}
-void emu_video_mode_change(int start_line, int line_count, int is_32cols)
+void emu_video_mode_change(int start_line, int line_count, int start_col, int col_count)
{
struct retro_system_av_info av_info;
vm_current_start_line = start_line;
vm_current_line_count = line_count;
- vm_current_is_32cols = is_32cols;
+ vm_current_start_col = start_col;
+ vm_current_col_count = col_count;
// 8bit renderes create a 328x256 CLUT image, while 16bit creates 320x240 RGB
vout_16bit = vout_format == PDF_RGB555 || (PicoIn.AHW & PAHW_32X);
vout_width = (vout_16bit ? VOUT_MAX_WIDTH : VOUT_8BIT_WIDTH);
vout_height = (vout_16bit ? VOUT_MAX_HEIGHT : VOUT_8BIT_HEIGHT);
vout_offset = (vout_16bit ? 0 : 8); // 8bit has 8 px overlap area on the left
- if (is_32cols) {
- // 256x240, with or w/o overlap on the left and 64 px on the right
- padding = (struct retro_hw_ps2_insets){start_line, vout_offset, vout_height - line_count - start_line, vout_width - 256.0f - vout_offset};
- } else {
- // 320x240, with or w/o overlap on the left and none on the right
- padding = (struct retro_hw_ps2_insets){start_line, vout_offset, vout_height - line_count - start_line, vout_width - 320.0f - vout_offset};
- }
+ padding = (struct retro_hw_ps2_insets){start_line, vout_offset, vout_height - line_count - start_line, vout_width - col_count - vout_offset};
int pxsz = (vout_16bit ? 2 : 1); // pixel size: RGB = 16 bits, CLUT = 8 bits
memset(vout_buf, 0, pxsz * vout_width * vout_height);
ps2->padding = padding;
}
#else
- vout_width = is_32cols ? VOUT_32COL_WIDTH : VOUT_MAX_WIDTH;
+ vout_width = col_count;
memset(vout_buf, 0, VOUT_MAX_WIDTH * VOUT_MAX_HEIGHT * 2);
if (vout_16bit)
PicoDrawSetOutBuf(vout_buf, vout_width * 2);
PicoDrawSetOutFormat(vout_format, 0);
if ((vm_current_start_line != -1) &&
(vm_current_line_count != -1) &&
- (vm_current_is_32cols != -1))
+ (vm_current_start_col != -1) &&
+ (vm_current_col_count != -1))
emu_video_mode_change(
- vm_current_start_line,
- vm_current_line_count,
- vm_current_is_32cols);
+ vm_current_start_line, vm_current_line_count,
+ vm_current_start_col, vm_current_col_count);
}
void lprintf(const char *fmt, ...)
{
if ((vm_current_start_line != -1) &&
(vm_current_line_count != -1) &&
- (vm_current_is_32cols != -1))
+ (vm_current_start_col != -1) &&
+ (vm_current_col_count != -1))
emu_video_mode_change(
- vm_current_start_line,
- vm_current_line_count,
- vm_current_is_32cols);
+ vm_current_start_line, vm_current_line_count,
+ vm_current_start_col, vm_current_col_count);
}
/* Reinitialise frameskipping, if required */
#include "../libpicofe/plat.h"\r
#include "../common/emu.h"\r
#include "../common/arm_utils.h"\r
+#include "../common/upscale.h"\r
#include "../common/version.h"\r
\r
#include <pico/pico_int.h>\r
const char *renderer_names32x[] = { "accurate", "faster", "fastest", NULL };\r
enum renderer_types { RT_16BIT, RT_8BIT_ACC, RT_8BIT_FAST, RT_COUNT };\r
\r
-static int out_x, out_y;\r
-static int out_w, out_h;\r
+static int out_x, out_y, out_w, out_h; // renderer output in render buffer\r
+static int screen_x, screen_y, screen_w, screen_h; // final render destination \r
+static int render_bg; // force 16bit mode for bg render\r
\r
void pemu_prep_defconfig(void)\r
{\r
}\r
\r
#define is_16bit_mode() \\r
- (currentConfig.renderer == RT_16BIT || (PicoIn.AHW & PAHW_32X))\r
+ (currentConfig.renderer == RT_16BIT || (PicoIn.AHW & PAHW_32X) || render_bg)\r
\r
static int get_renderer(void)\r
{\r
#undef p\r
}\r
\r
-static unsigned short *get_16bit_start(unsigned short *buf)\r
+/* render/screen buffer handling:\r
+ * In 16 bit mode, render output is directly placed in the screen buffer.\r
+ * SW scaling is handled in renderer (x) and in vscaling callbacks here (y).\r
+ * In 8 bit modes, output goes to the internal Draw2FB buffer in alternate\r
+ * renderer format (8 pix overscan at left/top/bottom), left aligned (DIS_32C).\r
+ * It is converted to 16 bit and SW scaled in pemu_finalize_frame.\r
+ *\r
+ * HW scaling always aligns the image to the left/top, since selecting an area\r
+ * for display isn't always possible.\r
+ */\r
+\r
+static u16 *screen_buffer(u16 *buf)\r
+{\r
+ // center the emulator display on the screen if screen is larger\r
+ if (currentConfig.scaling != EOPT_SCALE_HW)\r
+ buf += (g_screen_width-320)/2;\r
+ if (currentConfig.vscaling != EOPT_SCALE_HW)\r
+ buf += (g_screen_height-240)/2 * g_screen_ppitch;\r
+ return buf;\r
+}\r
+\r
+void screen_blit(u16 *pd, int pp, u8* ps, int ss, u16 *pal)\r
{\r
- // center the output on the screen\r
- int offs = (g_screen_height-240)/2 * g_screen_ppitch + (g_screen_width-320)/2;\r
- return buf + offs;\r
+ typedef void (*upscale_t)\r
+ (u16 *di,int ds, u8 *si,int ss, int w,int h, u16 *pal);\r
+ upscale_t upscale_hv[] = {\r
+ upscale_rgb_nn_x_4_5_y_16_17, upscale_rgb_snn_x_4_5_y_16_17,\r
+ upscale_rgb_bl2_x_4_5_y_16_17, upscale_rgb_bl4_x_4_5_y_16_17,\r
+ };\r
+ upscale_t upscale_h[] = {\r
+ upscale_rgb_nn_x_4_5, upscale_rgb_snn_x_4_5,\r
+ upscale_rgb_bl2_x_4_5, upscale_rgb_bl4_x_4_5,\r
+ };\r
+ upscale_t upscale_v[] = {\r
+ upscale_rgb_nn_y_16_17, upscale_rgb_snn_y_16_17,\r
+ upscale_rgb_bl2_y_16_17, upscale_rgb_bl4_y_16_17,\r
+ };\r
+ upscale_t *upscale;\r
+ int y;\r
+\r
+ // handle software upscaling\r
+ upscale = NULL;\r
+ if (currentConfig.scaling == EOPT_SCALE_SW && out_w == 256) {\r
+ if (currentConfig.vscaling == EOPT_SCALE_SW && out_h <= 224)\r
+ // h+v scaling\r
+ upscale = upscale_hv;\r
+ else\r
+ // h scaling\r
+ upscale = upscale_h;\r
+ } else if (currentConfig.vscaling == EOPT_SCALE_SW && out_h <= 224) {\r
+ // v scaling\r
+ upscale = upscale_v;\r
+ } else {\r
+ // no scaling\r
+ for (y = 0; y < out_h; y++)\r
+ h_copy(pd, pp, ps, 328, out_w, f_pal);\r
+ return;\r
+ }\r
+\r
+ upscale[currentConfig.filter & 0x3](pd, pp, ps, ss, out_w, out_h, pal);\r
}\r
\r
void pemu_finalize_frame(const char *fps, const char *notice)\r
{\r
if (!is_16bit_mode()) {\r
// convert the 8 bit CLUT output to 16 bit RGB\r
- unsigned short *pd = (unsigned short *)g_screen_ptr +\r
- out_y * g_screen_ppitch + out_x;\r
- unsigned char *ps = Pico.est.Draw2FB + 328*out_y + 8;\r
- unsigned short *pal = Pico.est.HighPal;\r
- int i, x;\r
+ u16 *pd = screen_buffer(g_screen_ptr) +\r
+ screen_y * g_screen_ppitch + screen_x;\r
+ u8 *ps = Pico.est.Draw2FB + 328*out_y + out_x + 8;\r
\r
- pd = get_16bit_start(pd);\r
PicoDrawUpdateHighPal();\r
- for (i = 0; i < out_h; i++, ps += 8) {\r
- for (x = 0; x < out_w; x++)\r
- *pd++ = pal[*ps++];\r
- pd += g_screen_ppitch - out_w;\r
- ps += 320 - out_w;\r
- }\r
+\r
+ screen_blit(pd, g_screen_ppitch, ps, 328, Pico.est.HighPal);\r
}\r
\r
if (notice)\r
void plat_video_set_buffer(void *buf)\r
{\r
if (is_16bit_mode())\r
- PicoDrawSetOutBuf(get_16bit_start(buf), g_screen_ppitch * 2);\r
+ PicoDrawSetOutBuf(screen_buffer(buf), g_screen_ppitch * 2);\r
}\r
\r
static void apply_renderer(void)\r
{\r
+ PicoIn.opt &= ~(POPT_ALT_RENDERER|POPT_EN_SOFTSCALE|POPT_DIS_32C_BORDER);\r
switch (get_renderer()) {\r
case RT_16BIT:\r
- PicoIn.opt &= ~POPT_ALT_RENDERER;\r
- PicoIn.opt &= ~POPT_DIS_32C_BORDER;\r
- PicoDrawSetOutFormat(PDF_RGB555, 0);\r
- PicoDrawSetOutBuf(get_16bit_start(g_screen_ptr), g_screen_ppitch * 2);\r
+ // 32X uses line mode for vscaling with accurate renderer, since\r
+ // the MD VDP layer must be unscaled and merging the scaled 32X\r
+ // image data will fail.\r
+ PicoDrawSetOutFormat(PDF_RGB555,\r
+ (PicoIn.AHW & PAHW_32X) && currentConfig.vscaling);\r
+ PicoDrawSetOutBuf(screen_buffer(g_screen_ptr), g_screen_ppitch * 2);\r
break;\r
case RT_8BIT_ACC:\r
- PicoIn.opt &= ~POPT_ALT_RENDERER;\r
- PicoIn.opt |= POPT_DIS_32C_BORDER;\r
+ // for simplification the 8 bit accurate renderer uses the same\r
+ // storage format as the fast renderer\r
PicoDrawSetOutFormat(PDF_8BIT, 0);\r
PicoDrawSetOutBuf(Pico.est.Draw2FB, 328);\r
break;\r
case RT_8BIT_FAST:\r
PicoIn.opt |= POPT_ALT_RENDERER;\r
- PicoIn.opt |= POPT_DIS_32C_BORDER;\r
PicoDrawSetOutFormat(PDF_NONE, 0);\r
break;\r
}\r
\r
if (PicoIn.AHW & PAHW_32X)\r
- PicoDrawSetOutBuf(get_16bit_start(g_screen_ptr), g_screen_ppitch * 2);\r
+ PicoDrawSetOutBuf(screen_buffer(g_screen_ptr), g_screen_ppitch * 2);\r
+ else if (is_16bit_mode()) {\r
+ if (currentConfig.scaling == EOPT_SCALE_SW) {\r
+ PicoIn.opt |= POPT_EN_SOFTSCALE;\r
+ PicoIn.filter = currentConfig.filter;\r
+ } else if (currentConfig.scaling == EOPT_SCALE_HW)\r
+ // hw scaling, render without any padding\r
+ PicoIn.opt |= POPT_DIS_32C_BORDER;\r
+ } else\r
+ PicoIn.opt |= POPT_DIS_32C_BORDER;\r
\r
Pico.m.dirtyPal = 1;\r
}\r
{\r
}\r
\r
+void pemu_sound_start(void)\r
+{\r
+ emu_sound_start();\r
+}\r
+\r
+void plat_debug_cat(char *str)\r
+{\r
+}\r
+\r
void pemu_forced_frame(int no_scale, int do_emu)\r
{\r
- unsigned short *pd = get_16bit_start(g_screen_ptr);\r
+ u16 *pd = screen_buffer(g_screen_ptr);\r
+ int hs = currentConfig.scaling, vs = currentConfig.vscaling;\r
\r
+ // create centered and sw scaled (if scaling enabled) 16 bit output\r
PicoIn.opt &= ~POPT_DIS_32C_BORDER;\r
- PicoDrawSetCallbacks(NULL, NULL);\r
Pico.m.dirtyPal = 1;\r
+ if (currentConfig.scaling) currentConfig.scaling = EOPT_SCALE_SW;\r
+ if (currentConfig.vscaling) currentConfig.vscaling = EOPT_SCALE_SW;\r
+ plat_video_set_size(320, 240);\r
\r
+ // render a frame in 16 bit mode\r
+ render_bg = 1;\r
emu_cmn_forced_frame(no_scale, do_emu, pd);\r
+ render_bg = 0;\r
\r
g_menubg_src_ptr = g_screen_ptr;\r
+ currentConfig.scaling = hs, currentConfig.vscaling = vs;\r
}\r
\r
-void pemu_sound_start(void)\r
+/* vertical sw scaling, 16 bit mode */\r
+static int vscale_state;\r
+\r
+static int cb_vscaling_begin(unsigned int line)\r
{\r
- emu_sound_start();\r
+ static int prevline = 999;\r
+\r
+ // at start of new frame?\r
+ if (line < prevline) {\r
+ // set y frame offset (see emu_change_video_mode)\r
+ u16 *dest = g_screen_ptr;\r
+ Pico.est.DrawLineDest = dest + screen_y * g_screen_ppitch;\r
+ vscale_state = 0;\r
+ }\r
+ prevline = line;\r
+ return 0;\r
}\r
\r
-void plat_debug_cat(char *str)\r
+static int cb_vscaling_nop(unsigned int line)\r
+{\r
+ return 0;\r
+}\r
+\r
+static int cb_vscaling_end(unsigned int line)\r
{\r
+ u16 *dest = Pico.est.DrawLineDest;\r
+ switch (currentConfig.filter) {\r
+ case 3: v_upscale_bl4_16_17(dest, g_screen_ppitch, 320, vscale_state);\r
+ break;\r
+ case 2: v_upscale_bl2_16_17(dest, g_screen_ppitch, 320, vscale_state);\r
+ break;\r
+ case 1: v_upscale_snn_16_17(dest, g_screen_ppitch, 320, vscale_state);\r
+ break;\r
+ default: v_upscale_nn_16_17(dest, g_screen_ppitch, 320, vscale_state);\r
+ break;\r
+ }\r
+ Pico.est.DrawLineDest = dest;\r
+ return 0;\r
}\r
\r
-void emu_video_mode_change(int start_line, int line_count, int is_32cols)\r
+void emu_video_mode_change(int start_line, int line_count, int start_col, int col_count)\r
{\r
+ // relative position in core fb and screen fb\r
+ out_y = start_line; out_x = start_col;\r
+ out_h = line_count; out_w = col_count;\r
+\r
+ PicoDrawSetCallbacks(NULL, NULL);\r
+ screen_x = screen_y = 0;\r
+ screen_w = 320, screen_h = 240;\r
+\r
+ switch (currentConfig.scaling) {\r
+ case EOPT_SCALE_HW:\r
+ screen_w = out_w;\r
+ break;\r
+ case EOPT_SCALE_NONE:\r
+ // center output in screen\r
+ screen_x = (screen_w - out_w)/2;\r
+ break;\r
+ }\r
+ switch (currentConfig.vscaling) {\r
+ case EOPT_SCALE_HW:\r
+ // NTSC always has 224 visible lines, anything smaller has bars\r
+ screen_h = (out_h < 224 ? 224 : out_h);\r
+ // handle vertical centering for 16 bit mode\r
+ screen_y = (screen_h - out_h) / 2;\r
+ if (is_16bit_mode())\r
+ PicoDrawSetCallbacks(cb_vscaling_begin, cb_vscaling_nop);\r
+ break;\r
+ case EOPT_SCALE_SW:\r
+ // NTSC always has 224 visible lines, anything smaller has bars\r
+ if (out_y > 7)\r
+ screen_y = out_y - 7;\r
+ // in 16 bit mode sw scaling is divided between core and platform\r
+ if (is_16bit_mode() && out_h < 240)\r
+ PicoDrawSetCallbacks(cb_vscaling_begin, cb_vscaling_end);\r
+ break;\r
+ case EOPT_SCALE_NONE:\r
+ // center output in screen\r
+ screen_y = (screen_h - out_h)/2;\r
+ break;\r
+ }\r
+\r
+ plat_video_set_size(screen_w, screen_h);\r
+ plat_video_set_buffer(g_screen_ptr);\r
+\r
// clear whole screen in all buffers\r
if (!is_16bit_mode())\r
memset32(Pico.est.Draw2FB, 0xe0e0e0e0, (320+8) * (8+240+8) / 4);\r
plat_video_clear_buffers();\r
-\r
- out_y = start_line; out_x = (is_32cols ? 32 : 0);\r
- out_h = line_count; out_w = (is_32cols ? 256:320);\r
}\r
\r
void pemu_loop_prep(void)\r
-#include "../libpicofe/gp2x/plat_gp2x.h"
-
// ------------ gfx options menu ------------
+static const char *men_scaling_opts[] = { "OFF", "software", "hardware", NULL };
+static const char *men_filter_opts[] = { "nearest", "smoother", "bilinear 1", "bilinear 2", NULL };
-const char *men_scaling_opts[] = { "OFF", "ON", NULL };
+static const char h_scale[] = "hardware scaling may not be working on some devices";
#define MENU_OPTIONS_GFX \
- // mee_enum ("screen scaling", MA_OPT_SCALING, currentConfig.scaling, men_scaling_opts), \
+ mee_enum_h ("Horizontal scaling", MA_OPT_SCALING, currentConfig.scaling, men_scaling_opts, h_scale), \
+ mee_enum_h ("Vertical scaling", MA_OPT_VSCALING, currentConfig.vscaling, men_scaling_opts, h_scale), \
+ mee_enum_h ("Scaler type", MA_OPT3_FILTERING, currentConfig.filter, men_filter_opts, NULL), \
#define MENU_OPTIONS_ADV
plat_video_flip();\r
}\r
\r
-void emu_video_mode_change(int start_line, int line_count, int is_32cols)\r
+void emu_video_mode_change(int start_line, int line_count, int start_col, int col_count)\r
{\r
int fb_w = 320, fb_h = 240, fb_left = 0, fb_right = 0, fb_top = 0, fb_bottom = 0;\r
\r
if (doing_bg_frame)\r
return;\r
\r
- if (is_32cols) {\r
- fb_w = 256;\r
- fb_left = fb_right = 32;\r
- }\r
+ fb_w = col_count;\r
+ fb_left = start_col;\r
+ fb_right = 320 - (fb_w+fb_left);;\r
\r
switch (currentConfig.scaling) {\r
case SCALE_1x1:\r
g_layer_h = fb_h;\r
break;\r
case SCALE_2x2_3x2:\r
- g_layer_w = fb_w * (is_32cols ? 3 : 2);\r
+ g_layer_w = fb_w * (col_count < 320 ? 3 : 2);\r
g_layer_h = fb_h * 2;\r
break;\r
case SCALE_2x2_2x2:\r
fb_h = line_count;\r
break;\r
}\r
- g_osd_fps_x = is_32cols ? 232 : 264;\r
+ g_osd_fps_x = col_count < 320 ? 232 : 264;\r
g_osd_y = fb_top + fb_h - 8;\r
\r
pnd_setup_layer(1, g_layer_x, g_layer_y, g_layer_w, g_layer_h);\r
}
/* prepare for MD screen mode change */
-void emu_video_mode_change(int start_line, int line_count, int is_32cols)
+void emu_video_mode_change(int start_line, int line_count, int start_col, int col_count)
{
- h32_mode = is_32cols;
- out_y = start_line; out_x = (is_32cols ? 32 : 0);
- out_h = line_count; out_w = (is_32cols ? 256:320);
+ h32_mode = col_count < 320;
+ out_y = start_line; out_x = (h32_mode ? 32 : 0);
+ out_h = line_count; out_w = (h32_mode ? 256:320);
vidResetMode();
if (h32_mode) // clear borders from h40 remnants
PicoDrawSetOutFormat(PDF_RGB555, 1);
}
-void emu_video_mode_change(int start_line, int line_count, int is_32cols)
+void emu_video_mode_change(int start_line, int line_count, int start_col, int col_count)
{
- EmuScreenRect.left = is_32cols ? 32 : 0;
- EmuScreenRect.right = is_32cols ? 256+32 : 320;
+ EmuScreenRect.left = start_col;
+ EmuScreenRect.right = start_col + col_count;
EmuScreenRect.top = start_line;
EmuScreenRect.bottom = start_line + line_count;
get_define OFS_Pico_ Pico romsize ; echo "$line" >>$fn
get_define OFS_Pico_ Pico est ; echo "$line" >>$fn
+get_define OFS_PicoIn_ PicoInterface opt ; echo "$line" >>$fn
+get_define OFS_PicoIn_ PicoInterface filter ; echo "$line" >>$fn
+
get_define OFS_EST_ PicoEState DrawScanline ; echo "$line" >>$fn
get_define OFS_EST_ PicoEState rendstatus ; echo "$line" >>$fn
get_define OFS_EST_ PicoEState DrawLineDest ; echo "$line" >>$fn
+get_define OFS_EST_ PicoEState DrawLineDestIncr ; echo "$line" >>$fn
get_define OFS_EST_ PicoEState HighCol ; echo "$line" >>$fn
get_define OFS_EST_ PicoEState HighPreSpr ; echo "$line" >>$fn
get_define OFS_EST_ PicoEState Pico ; echo "$line" >>$fn
notaz.gp2x.de / picodrive.git / commitdiff